Journal of Artificial Intelligence and Engineering Applications (JAIEA)

Library Information System Design at State Senior High School 5 Prabumulih

2025-08-10T23:37:05+07:00

Information systems using very sophisticated and modern computer technology will be very helpful and facilitate us in processing data that can save time, space and costs. Based on the existing problems, this study aims to design a web-based library information system at SMA Negeri 5 Prabumulih This system development method uses a research approach with SDLC or Software Development Life Cycle. While the model used in the design of this library information system is the agile model. The model provides a sequential or sequential software lifeflow approach starting from analysis, design, coding and testing. The programming language itself uses PHP with a codeigniter framework and for the database it uses MySQL/XAMPP. The results of this study show that the library information system at SMA Negeri 5 prabumulih is included in the feasible category after testing using black box testing category.

Design and Construction of Academic Information System Website at PKBM As-salaam using the RAD Method

2025-08-14T10:08:04+07:00

This research aims to design and develop a web-based information system for Community Learning Activity Center As-Salaam using the Rapid Application Development method. The purpose is based on the fact that the current administrative processes are still carried out manually, either through printed forms or using Microsoft Word, which poses risks of input errors in student data and difficulties in delivering class schedule information to relevant parties. The proposed information system is designed to improve efficiency and accuracy in administrative management, including student registration, class scheduling, and student data processing. The RAD approach is chosen because it enables fast and iterative system development by involving users in every development stage, ensuring that the system is responsive to real user needs. The system will be developed using PHP as the programming language and MySQL as the database. The implementation is expected to facilitate administrative processes, improve data accuracy, accelerate data-driven decision-making, and provide more transparent and efficient services. Ultimately, PKBM As-Salaam is expected to deliver better and more modern non-formal education services.

Development of the Rad Method for Administrative Service Information Systems (Case Study: Jiwa Baru Village)

2025-08-14T15:43:14+07:00

Administrative services in Jiwa Baru Village are still handled manually, causing delays, data errors, and inefficiency. This study aims to develop a web-based administrative information system using the Rapid Application Development (RAD) method. RAD was chosen for its speed and iterative approach involving active user participation. The system includes features for document requests and digital archiving. Data were collected through interviews and observations in Jiwa Baru Village. The system was evaluated using black-box testing. The results show that the system improves service efficiency, reduces errors, and provides easier data access for both village officials and residents. This system supports more modern and transparent public service delivery.

Design of Academic Information System at Bintang Harapan Preschool

2025-08-14T22:52:09+07:00

PAUD Bintang Harapan is an early childhood education located in Talang Batu Village, Rambang Kapak Tengah District, Prabumulih City, South Sumatra Province. Currently, PAUD Bintang Harapan in storing academic data is still in the form of archives, both student attendance, report card filling and tuition payment reports, so it is difficult to search for data. Therefore, the author tries to create a website-based academic information system using PHP and MySQL and UML assistance in designing the system created. This design uses the Extreme programming method, so it can be concluded that this system is feasible if the academic information system is implemented at PAUD Bintang Harapan.

Experimental Evaluation and Performance Analysis of a Dual-Factor IoT-Based Smart Door Security System

2025-08-14T10:13:41+07:00

This research presents the design, implementation, and experimental evaluation of an Internet of Things (IoT)-enabled smart door security system integrating biometric and PIN-based authentication. The study benchmarks system performance through quantitative metrics including authentication accuracy, latency, and wireless communication stability. Unlike prior single-factor systems, the proposed dual-factor model demonstrates improved resistance to unauthorized access and enhanced reliability under varied environmental conditions. Comparative analysis with traditional and commercial smart locks reveals that the developed system achieves faster response times (<2 s), higher authentication accuracy (98.5%), and lower cost while maintaining user convenience. The findings contribute an empirically validated framework for secure, modular, and cost-efficient IoT-based access control.

Designing E-Voting to Increase the Efficiency and Transparency of the Election of the OSIS Chairman at SMP Negeri 7 Prabumulih

2025-08-18T17:07:09+07:00

Cindy Feronica Wulandari (2021210014), Supervisor I Mr. Andi Christian S.kom., M.kom., and Supervisor II Ms. Nur Aini Hutagalung S.kom.M.,Si.,M.kom., entitled "Designing E-Voting Technology to Increase the Efficiency and Transparency of the Student Council President Election at SMP Negeri 7 Prabumulih." This research aims to design an Android-based e-voting application to increase the efficiency and transparency of the Student Council President election at SMP Negeri 7 Prabumulih. The manual voting process currently used has resulted in various obstacles such as counting errors, damaged ballots, and is time-consuming. Using the Rapid Application Development (RAD) method, this application was developed to make the voting and vote counting process faster, more accurate, and more secure, while encouraging active student participation in the school's democratic process. Therefore, it can be concluded that the Android-based e-voting application can increase the efficiency, transparency, and participation in the Student Council President election at SMP Negeri 7 Prabumulih, as well as simplify the digital voting and vote counting process.

Design of a Web-Based Application for Business Enrollment Number (NIB) Recording at the Prabumulih City Dpmptsp

2025-08-18T21:39:22+07:00

This proposal aims to design and develop a web-based application for recording and reporting Business Identification Numbers (NIB) at the Investment and One-Stop Integrated Service (DPMPTSP) of Prabumulih City. Currently, the process of recording NIBs is done manually using a report book and Microsoft Excel, which can lead to errors, inefficiencies, and difficulty in retrieving and reporting data. The web-based application to be developed will replace the manual system with a more efficient, accurate, and easily accessible data management system. Equipped with automated reporting features and a dashboard to monitor NIB status in real-time, this application is expected to accelerate administrative processes, minimize data input errors, and provide ease for business owners to access information about their NIB status in a transparent and timely manner.

An Analysis of User Satisfaction Toward the Maxim Application in Medan City using the E-Servqual Model

2025-09-08T11:09:30+07:00

The development of information technology has facilitated various activities, including online transportation services. Maxim, as one of the digital transportation service providers, has received numerous user complaints, particularly regarding its navigation features, interface design, and electronic payment system. This study aims to analyze user satisfaction with the Maxim application in Medan City using the E-Servqual method, which includes five dimensions: tangibles, reliability, responsiveness, assurance, and empathy. The research used a quantitative approach by distributing online questionnaires to 104 respondents. The results showed that all dimensions had negative GAP values, with tangibles being the most dominant factor affecting user satisfaction. As a solution, a real-time dashboard prototype was designed to help developers monitor feature evaluations directly. This research is expected to serve as a reference for improving the quality of Maxim’s services in the future.

Classification of Purple Passion Fruit Ripeness Levels Using Convolutional Neural Network (CNN)

2025-10-27T08:40:32+07:00

Passiflora edulis Sims (purple passion fruit) is a fruit that offers numerous health benefits and possesses high economic value. However, the manual assessment of ripeness by traders tends to be subjective and inconsistent, leading to post-harvest losses of up to 50%. This study developed a classification model for determining the ripeness level of purple passion fruit using a Convolutional Neural Network (CNN) and implemented it in a web-based application. The CNN model was designed to classify four ripeness stages (unripe, half-ripe, ripe, and rotten) with the addition of a non-passion-fruit class to enhance the system’s robustness. The dataset consisted of 2,000 images divided into five classes: four ripeness levels of purple passion fruit (unripe, half-ripe, ripe, and rotten) and one non-passion-fruit class as a comparator. All images were in JPG and PNG formats. The CNN architecture comprised four convolutional layers with 16, 32, 64, and 128 filters, respectively. Evaluation of various data-splitting ratios (80:20, 70:30, 60:40) and learning rates (0.001, 0.0001, 0.01) showed that the optimal configuration was achieved at a ratio of 80:20 with a learning rate of 0.001, resulting in a training accuracy of 96.72% and a testing accuracy of 95.76%, with a loss value of 0.1811. Validation using 5-Fold Cross Validation produced an average accuracy of 95.40%. The model was integrated into a web application developed using Flask and JavaScript, deployed on the PythonAnywhere cloud platform, enabling users to upload images and automatically obtain ripeness predictions to assist traders in sorting fruits more quickly and accurately.

Design of a Modern Web-Based Mail Management Information System Using the Prototype Model at PT Kalimantan Teknologi Indonesia

2025-10-31T15:46:56+07:00

PT Kalimantan Teknologi Indonesia manages its incoming and outgoing mail using a conventional manual approach, which often results in inefficiency, loss of documents, and difficulties in information retrieval. This study aims to design a web-based mail management information system that enhances the efficiency, accuracy, and traceability of administrative correspondence. The research adopts the Prototype development methodology, consisting of requirement analysis, system design, prototype creation, and evaluation. Data were collected through interviews, observations, and document analysis to identify user needs and administrative workflows. The system was developed using Figma employed for interface prototyping. The resulting system provides features for digital recording, classification, searching, and reporting of both incoming and outgoing mail. Testing results indicate that the prototype effectively reduces administrative workload, improves data accuracy, and accelerates document tracking. This study demonstrates that a well-structured digital correspondence system can serve as a foundation for administrative transformation in small and medium-sized enterprises.

Optimization of Financial Management at Idhotun Nasyi’in Islamic Boarding School using a Website Application

2025-11-03T17:40:35+07:00

Idhotun Nasyi'in Islamic Boarding School faces challenges in financial management due to its manual reliance on Microsoft Excel, which is prone to data loss, input errors, and tracking difficulties. This research aims to design and build a web-based financial management application to address these issues. Developed using the System Development Life Cycle (SDLC) waterfall model with PHP and MySQL, the application features core functionalities such as income and expense management, transaction categorization, and real-time financial report generation in PDF format. Black box testing results indicate that the application functions effectively, simplifying the tasks of administrators and treasurers in monitoring cash flow. The implementation of this system is expected to enhance the accuracy, efficiency, and transparency of the boarding school's financial management.

Brute-Force Attack Detection on Computer Networks Using Artificial Neural Network

2025-11-04T16:42:11+07:00

This research aims to develop a brute-force attack detection system on computer networks using the Artificial Neural Network (ANN) algorithm. This security problem is crucial, especially in the banking sector because it can threaten login systems and sensitive customer data. The research methods include data cleansing, feature selection using the Wrapper method, ANN model training, and performance evaluation using datasets from Kaggle which include four classes of network traffic, namely Normal, Brute-force FTP, Brute-force SSH, and Web Attack Brute-force. The test results showed that the ANN model achieved an accuracy of 95%, precision of 91%, and the best performance in the Brute-force FTP class with an accuracy of 98.3%. This system has proven to be effective in detecting brute-force attack patterns and can improve the security of banking networks adaptively. This research broadens the insights of the application of ANN in network security and provides a basis for the development of systems that are more responsive to cyber threats.

Analysis of the Use of Learning Media in English Learning in the Kurikulum Merdeka at SMA Negeri 1 Idanogawo

2025-11-05T13:59:04+07:00

This study investigates the use of learning media in English learning media under the Kurikulum Merdeka at SMA Negeri 1 Idanogawo using a descriptive qualitative method. Data were collected through observations and interviews with English teachers, alongside observations of 11th grade students.The findings reveal that the employment of both digital and conventionalmedia effectively boosts student engagement, motivation, and comprehension. This media utilization is crucial for adapting to the diverse characteristics and needs of the students. However, teachers face notable challenges, including limited technological facilities, inadequate school infrastructur, and insufficient time for designing and implementing innovative media. The choice of learning media is influenced by factors such as facility availability, ease of use, relevance to the material, and suitability for students characteristics. The study underscores the critical need for support from both the school and the government, specifically in the form of training, infrastructure provision, and profesional development, to optimize media usage. Furthermore, it highlights the essential role of teacher creativity in selecting and developing media appropriate for the Kurikulum Merdeka, making learning more effective and adaptive to 21st century demands. Ultimately, the appropriate and innovative use of learnig media can significantly improve the quality of English learning and enhance student competencies at SMA Negeri 1 Idanogawo.

Spam Message Classification Using the Naïve Bayes Algorithm Based on RapidMiner

2025-11-06T21:17:52+07:00

This study implements the Naïve Bayes algorithm for classifying spam and non-spam (ham) messages using the RapidMiner Studio platform. The dataset used was obtained from the SMS Spam Collection Dataset on the Kaggle platform, which consists of 5,759 messages with a distribution of 4,075 ham messages and 1,291 spam messages. The research stages included text pre-processing, model training, and performance evaluation using accuracy, precision, recall, and F1-score metrics. The experimental results showed that the Naïve Bayes model achieved an accuracy of 89.64% with a precision of 56.93%, a recall of 100%, and an F1-score of 72.56%. The research findings indicate that the Naïve Bayes algorithm is effective in detecting spam messages with adequate accuracy, and prove that RapidMiner is an efficient tool for implementing machine learning methods in text classification.

Clustering Provinces in Indonesia Based on Economic Indicators Using the K-Means Algorithm

2025-11-07T10:53:57+07:00

This study aims to analyze and classify the level of economic development in provinces in Indonesia using the K-Means algorithm. The data used includes three main indicators, namely Gross Regional Domestic Product (GRDP) per capita, percentage of poor population, and Human Development Index (HDI) in 2024 obtained from the Central Statistics Agency (BPS). The data was processed through normalization and analysis using the Elbow method to determine the optimal number of clusters. The results were evaluated using the Davies–Bouldin Index (DBI) to assess the level of separation and compactness between clusters. The results show that the most effective division consists of three groups representing high, medium, and low levels of development. Provinces such as DKI Jakarta and Riau are included in the high development cluster, Central Java and South Sulawesi are in the medium cluster, while Papua and East Nusa Tenggara are in the low cluster. These results show that machine learning methods, particularly K-Means, are capable of identifying patterns of regional economic inequality and provide a useful basis for the government in formulating more targeted and equitable development policies.

Automated Diagnosis Assistant with Random Forest Medical Image and Algorithm Feature Extraction

2025-11-09T20:34:01+07:00

Medical image-based disease diagnosis is a complex process and requires a high level of expertise. This study aims to develop an Automatic Diagnosis Assistant using a combination of image feature extraction techniques and Random Forest (RF) classification algorithms. Medical images are processed to extract meaningful textural features, such as using the Gray Level Co-occurrence Matrix (GLCM), which is then used to train the RF model. To address the problem of data imbalance that is common in medical datasets, the SMOTE technique is applied. The performance of the model is evaluated and optimized using Randomized Search to find the best hyperparameters. The results showed that the optimized RF model was able to achieve high accuracy, with significant improvements in the Recall and F1-Score metrics compared to the baseline model. This automated diagnostic assistant is expected to be an effective tool for medical personnel in speeding up and improving diagnostic accuracy, especially in cases with high image volumes.

Evaluation of Machine Learning Algorithms in Sentiment Analysis of the Satu Sehat Application

2025-11-10T15:02:04+07:00

This study aims to analyze and compare the performance of three sentiment classification algorithms—Support Vector Machine (SVM), Naïve Bayes (NB), and K-Nearest Neighbor (K-NN)—in classifying user reviews of the Satu Sehat application. The data preprocessing stage involves several steps, including text cleaning through normalization, removal of punctuation, numbers, and irrelevant characters, as well as the elimination of stopwords. Subsequently, stemming is performed to reduce words to their root forms. Feature extraction is conducted using the CountVectorizer method with a bag-of-words approach, which converts textual data into numerical representations. The dataset is then divided into training and testing subsets using an 80:20 train-test split ratio. Model performance is evaluated through a confusion matrix, producing key evaluation metrics such as accuracy, precision, recall, and F1-score. Based on the results of testing 9,192 user reviews, the SVM algorithm with a linear kernel demonstrated the best overall performance compared to NB and K-NN, as indicated by the highest accuracy score. These findings suggest that SVM is more effective in handling high-dimensional textual features, making it a highly suitable algorithm for sentiment analysis of digital health application reviews, particularly those related to Satu Sehat.

Geospatial Analysis of Global Temperature and Humidity Variations Using Integrated Meteorological Data

2025-11-11T11:10:56+07:00

Global climate monitoring is crucial for understanding variations in temperature and humidity, which directly influence ecosystems, human health, and socio-economic activities. This study presents a Geographic Information System (GIS)-based analysis and visualization of global temperature and humidity patterns using historical hourly weather data from 2012 to 2017. The dataset, obtained from open-access sources, was processed and analyzed in Google Colab using Python libraries such as pandas, geopandas, folium, and plotly. Data preprocessing involved merging city-level observations, cleaning missing values, and calculating mean temperature and humidity per location. The resulting dataset was then visualized through an interactive global map and a scatter plot to identify spatial relationships between the two climatic variables.To quantify these spatial relationships, a statistical correlation analysis was conducted, revealing a weak negative relationship between temperature and humidity (r = -0.25) across global regions.The findings reveal that regions near the equator exhibit consistently high temperatures and humidity, while higher-latitude cities show lower temperatures and more variable moisture levels. This GIS-based approach demonstrates the potential of open meteorological data for climate pattern recognition and supports reproducible workflows for environmental analysis. The results highlight the importance of integrating data science tools with GIS for accessible and scalable global climate visualization.

Baby Supplies Sales Prediction System using the Single Exponential Smoothing Method at Little Queen Baby Shop

2025-11-11T16:03:18+07:00

The increasing demand for baby equipment in Indonesia in recent years has created significant business opportunities for the retail sector, including Little Queen Baby Shop. However, seasonal fluctuations in demand often lead to stock management problems such as overstock and out of stock, which affect storage costs and customer satisfaction. This research aims to design and develop a sales prediction system for baby products using the Single Exponential Smoothing (SES) method as a solution to minimize forecasting errors and support data-driven decision-making. The research method involved collecting secondary sales data from January to November 2024, which was then processed using the SES algorithm with a smoothing parameter (α) to determine the optimal prediction values with the lowest error rate. The system was developed as a web-based application using PHP programming language and MySQL database, equipped with features such as transaction recording, stock management, sales analysis, and prediction reports for upcoming periods. The implementation results show that the SES-based prediction system provides sufficiently accurate forecasts, as indicated by low values of Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Mean Squared Error (MSE). This system enables Little Queen Baby Shop to optimize stock management, reduce the risk of losses due to excessive or insufficient inventory, and improve both operational efficiency and customer satisfaction.

Sentiment Analysis of Indonesian National Team Failure in the 2026 World Cup Qualifications Using Support Vector Machine Algorithm

2025-11-13T19:45:53+07:00

The Indonesian National Team's failure in the 2026 World Cup qualifiers has generated diverse responses on social media, particularly on Ferry Irwandi's YouTube channel. This study aims to analyze public sentiment towards the national team's performance based on YouTube user comments. The method used is a Support Vector Machine (SVM) with stages of data scraping, pre-processing (cleaning, case folding, normalization, tokenization, stopword removal, stemming), lexicon-based automatic labeling, and model evaluation using a confusion matrix. The data consists of 8,353 comments divided with a ratio of 80:20 for training and testing. The results show that the SVM algorithm is able to classify comments into two classes, positive and negative, with an accuracy of 81%, a precision of 82%, a recall of 83%, and an F1-score of 82%. These results demonstrate the effectiveness of SVM in accurately and stably identifying public opinion towards the Indonesian National Team's failure.

Development of Environmental Cleanliness Education Game for Grade 5 Students at SD Inpres Kalu

2025-11-15T19:40:13+07:00

Technological advances have brought major changes in various aspects of life, including the world of education. One form of use of technology in education is the development of interactive learning media such as educational games. This research aims to develop an Android-based educational game that raises the theme of environmental cleanliness and is intended for 5th grade students of SD Inpres Kalu, East Sumba. The background of this research is based on the low understanding of students on the importance of maintaining environmental cleanliness, which is caused by conventional learning methods that are less interesting and less interactive. The educational game developed will contain materials such as types of waste, how to sort and dispose of waste, and dirty environmental impacts. The research method used is research and development (R&D) with a waterfall model that includes the stages of needs analysis, design, implementation, verification, and maintenance. Supporting data were obtained through interviews, observations, and literature studies. The results of the trial showed an increase in students' understanding of environmental hygiene materials, which was evidenced by an increase in the average score from 78.0 in the pre-test to 87.2 in the post-test, with a difference of 9.2 points or an increase of 11.79%. Testing using the Black Box Testing method showed that all in-game features performed as intended, while the System Usability Scale (SUS) test results obtained an average score of 83.5, which is in the excellent category.

Implementation of Web Based Motorcycle Sales Prediction System Using the Least Squares Method

2025-11-16T06:33:24+07:00

The development of information technology has brought significant changes in business data management, including in the automotive industry. Dony Jaya Motor, as one of the motorcycle dealers, faces challenges in predicting sales, particularly in balancing stock availability with market demand. This study aims to develop a web-based motorcycle sales prediction system using the Least Squares method due to its ability to identify linear trend patterns from historical data, producing accurate and measurable sales projections. The data used cover motorcycle sales from May 2024 to April 2025. The implementation results show that the Least Squares method provides good predictive accuracy, with the average Mean Absolute Percentage Error (MAPE) value below 10%, indicating a very low prediction error rate. For example, for the Honda Beat 2015 type, the predicted sales for May 2025 were 5.67 units compared to the actual 6 units, resulting in a MAPE value of 4.67%. The developed system includes features for data input, graphical visualization, and real-time prediction reporting. The application of the Least Squares method in this web-based system has proven to assist management in stock planning, improve decision-making processes, and enhance overall operational efficiency and effectiveness within the company.

Implementation of the Content-Based Filtering Method in Menu Recommendations at Pandawa Pondok Kopi

2025-11-19T07:05:19+07:00

The rapid growth of the coffee shop industry and the wide variety of menu offerings at Pandawa Pondok Kopi demand a system capable of delivering accurate and personalized menu recommendations. This study aimed to develop a web based menu recommendation application using Content Based Filtering (CBF), leveraging TF-IDF for document vectorization and Cosine Similarity to measure product description similarity.The system was implemented with PHP and MySQL, featuring a responsive interface across three main modules: the homepage (displaying the menu list), the menu detail page (providing full information and similar recommendations), and the admin dashboard (for menu data management). Menu descriptions were preprocessed (tokenization, stop word removal, and stemming) before computing TF-IDF weights. Given a user’s selected menu item, the system calculated Cosine Similarity between its description vector and those of all other menu items, then presents the top three matches. Functionality was verified via Black Box Testing to ensure that admin login, menu addition/editing, recommendation displays, and interface navigation conform to specifications. Test results showed an average Cosine Similarity score ranging from 0.62 to 0.78, indicating satisfactory accuracy in matching user preferences. The system also achieved an average response time of under one second under standard load, meeting efficiency criteria.In conclusion, the Content Based Filtering implementation successfully enhances the relevance of menu recommendations and user experience, thereby supporting increased customer satisfaction and operational effectiveness at Pandawa Pondok Kopi.

Sentiment Analysis of Mobile Legends Game using Naïve Bayes, K-Nearest Neighbors and Support Vector Machine Algorithm

2025-11-19T13:49:58+07:00

Sentiment analysis of Mobile Legends: Bang Bang (MLBB) user reviews is very important for understanding public satisfaction and perspectives. Therefore, this study aims to analyze and compare the performance of three Machine Learning algorithms: Naïve Bayes (NB), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM) in classifying user review sentiments. A supervised machine learning approach was applied using 6,000 reviews obtained from a secondary Kaggle dataset, involving Data Preprocessing and Feature Extraction (TF-IDF) stages, followed by an 80:20 Data Split for model training. The comparison of metric results shows that the Support Vector Machine (SVM) model provides the best overall performance, achieving 79.88% Accuracy and 78.06% F1-Score, although NB slightly outperforms in the Precision metric. In conclusion, SVM's performance proves this algorithm is superior in classifying Indonesian-language mobile game review sentiments, providing strategic insights for MLBB developers in making service improvement decisions.

Web-Based Spare Parts Expenditure Recording Portal with Read-Only Pull from ERP Infor (PT CBI Case Study)

2025-11-20T12:17:34+07:00

A companion portal for spare parts management was developed at the PT Century Batteries Indonesia (PT CBI) warehouse to address manual record keeping that is prone to discrepancies and difficult to trace. This article focuses on two key aspects: a checkout flow that validates expenditure amounts against Portal stock, and a reconciliation mechanism that pulls on-hand data from Infor's ERP read-only system to maintain ERP data integrity. The system was developed following the Extreme Programming (XP) methodology with short iterations and regular feedback from warehouse users. Black-box testing and UAT (User-to-User) testing demonstrated that the main flow functioned as expected; all assessed features were accepted with scores of 4.1–4.8. Consequently, discrepancies were detected more quickly and addressed through adjustments in Infor; the Portal then pulled back on-hand to ensure consistency. These results demonstrate that the "checkout in Portal → adjustment in Infor → pulled back on-hand (read-only)" pattern effectively reduces errors caused by manual recording while improving transaction traceability in the spare parts expenditure process in the warehouse environment.

Keywords : warehouse management; spare part checkout; Infor ERP; read-only integration; Extreme Programming;

Application of Support Vector Machine for Classification of Toddlers Nutritional Status Based on Anthropometric Data

2025-11-20T15:19:50+07:00

Stunting remains a major health issue in Indonesia, especially among toddlers. This study aims to classify the nutritional status of toddlers (stunted and non-stunted) using anthropometric data from the Kaggle public dataset with the Support Vector Machine (SVM) algorithm. This dataset includes data on the height, weight, age, and gender of toddlers. It should be emphasized that the data does not originate from the Ciherang Bandung Posyandu, but rather the Posyandu is used only as a context for the potential application of the developed model. The process includes data acquisition, preprocessing (including normalization and data balancing using SMOTE), SVM model training, and evaluation with accuracy, precision, recall, F1-score, and ROC-AUC. The model was trained with an 70:30 data split and optimal parameters (C=1.0, gamma=0.01, kernel=RBF). The results showed high performance, indicating that this model can support early detection of stunting and the implementation of decision support systems in public health services.

Comparative Performance Analysis of Multilayer Perceptron and Long Short-Term Memory for Daily Demand Forecasting in E-Commerce Delivery Platforms

2025-11-21T19:09:12+07:00

This study compares the performance of two deep learning architectures—Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM)—for daily demand forecasting on an e-commerce delivery platform. The dataset consists of 1,827 daily observations from 2020 to 2024 and includes operational, temporal, and behavioral features such as holiday indicators, promotion signals, active customers, and delivery time. Data preprocessing includes cleaning, feature engineering, scaling, and sequence generation using a 30-day sliding window. Both models were trained and evaluated using consistent experimental settings and performance metrics. The results show that the LSTM model achieves better accuracy than the MLP model, with an RMSE of 811.81 compared to 830.15, while the difference in MAE between the two models remains minimal. LSTM demonstrates superior capability in capturing temporal dependencies and reacting to rapid demand fluctuations, whereas both models face challenges when predicting sudden demand spikes. These findings indicate that memory-based models such as LSTM are more effective for highly volatile time-series forecasting in e-commerce operations. However, performance can be further improved with the addition of external variables such as real-time promotions, weather conditions, and multivariate features.

FP-Growth for Data-Driven Purchase Pattern Analysis and Product Recommendations at Flanetqueen Store

2025-11-21T14:38:43+07:00

The advancement of information technology has encouraged the use of data analytics to support data-driven business decision-making. This study aims to analyze purchasing patterns of hoodie products and provide product recommendations for customers at Flanetqueen Store using the FP-Growth (Frequent Pattern Growth) algorithm. The research applies the Knowledge Discovery in Database (KDD) framework, consisting of five stages: data selection, preprocessing, transformation, data mining, and interpretation/evaluation. The dataset comprises hoodie sales transactions recorded from January to December 2024. Data analysis was conducted using RapidMiner Studio version 10.3 with a minimum support of 0.2 and minimum confidence of 0.4. The analysis produced 26 itemsets and 11 association rules indicating product correlations. The strongest rule, Bloods → Champion, achieved a confidence of 0.414, revealing that customers who purchased Bloods hoodies were also likely to buy Champion hoodies. These findings were used to design cross-selling strategies and generate relevant product recommendations. The study demonstrates that FP-Growth effectively extracts frequent purchase patterns and contributes to the development of data-driven recommendation systems in the local fashion retail industry.

Application of the K-Means Algorithm in the Segmentation of 3kg Lpg Customers

2025-11-24T14:08:44+07:00

This research was motivated by PT Sumber Perkasa Mandiri's need to understand the purchasing patterns of 3 kg LPG gas customers more accurately in order to improve the effectiveness of its marketing strategy. The purpose of this study was to apply the K-Means Clustering algorithm to form customer segmentation based on transaction behavior. The method used is a quantitative approach with sales data analysis of 850 records through the stages of data selection, preprocessing, attribute transformation, and modeling using RapidMiner Studio. Model evaluation was carried out using the Davies-Bouldin Index to determine the optimal number of clusters. The results of the study show the formation of two main clusters, namely the premium customer cluster with high purchase frequency and high loyalty, and the low-activity customer cluster that only makes purchases when necessary. The best DBI value at K=2 of 0.057 indicates excellent cluster separation quality. These findings conclude that K-Means Clustering is effective in identifying differences in consumption behavior, and its implications provide a strategic basis for companies to design loyalty programs for high-value customers and more intensive promotions for low-activity customers.

Application of Decision Tree Algorithms to Classify the Sales Results of Kangen Kripik Sme Products

2025-11-24T14:08:08+07:00

Micro, Small, and Medium Enterprises (MSMEs) play a vital role in strengthening the national economy; however, many still face challenges in managing and analyzing sales data effectively. This study aims to classify product sales results at UMKM Kangen Kripik Mang Acep by applying the Decision Tree algorithm as a data classification method based on machine learning. A quantitative experimental approach was employed to evaluate the model’s performance using one-year sales data, including attributes such as product variants, sales volume, sales channels, and marketing regions. Data processing was conducted using RapidMiner software following the Knowledge Discovery in Databases (KDD) framework, which includes data selection, preprocessing, transformation, data mining, and model evaluation. The results indicate that the Decision Tree algorithm successfully classified sales regions (Garut, Bandung, and Sumedang) with an accuracy rate of 96.48%, identifying “Units Sold (pcs)” as the most influential attribute for distinguishing marketing areas. These findings demonstrate that the Decision Tree method is not only effective in improving data analysis efficiency but also provides valuable strategic insights for data-driven business decision-making in MSMEs

Literature Review: Transitioning usage from BFS and DFS to Heuristic Search in the Modern AI Era

2025-11-26T20:37:15+07:00

Uninformed search algorithms, specifically Breadth-First Search (BFS) and Depth-First Search (DFS), encounter significant scalability limitations when addressing complex problem spaces in modern Artificial Intelligence (AI) ecosystems. This study investigates the paradigm shift toward intelligent heuristic algorithms through a systematic literature review and comparative analysis of 24 recent academic sources. The evaluation focuses on three primary domains: logical problem solving, robotic navigation, and data infrastructure management. Results demonstrate that heuristic methods, such as A-Star and hybrid variants like PrunedBFS, offer superior time efficiency and memory optimization for autonomous navigation and massive computing tasks. Nevertheless, classic algorithms retain functional relevance for specific scenarios requiring exhaustive exploration. Furthermore, this study reveals that algorithmic evolution has fundamentally transformed digital infrastructure, driving a shift from Search Engine Optimization (SEO) to Answer Engine Optimization (AEO) and necessitating adaptive cybersecurity architectures. The research concludes that the future of AI development relies not on substitution, but on a collaborative synthesis integrating the robustness of classic methods with the adaptability of modern heuristics.

Comparison of Memory Efficiency and Computation Time of Bubble Sort, Insertion Sort, and Intro Sort Algorithms Using the C++ Programming Language

2025-11-27T12:43:15+07:00

Data sorting is a fundamental step in the computer process that greatly affects the effectiveness of programs and overall system performance. In this study, three sorting algorithms, namely Bubble Sort, Insertion Sort, and Intro Sort, are analyzed and compared using recursive and iterative approaches. Bubble Sort serves as a basic algorithm example to understand the basic idea of element exchange, while Insertion Sort was chosen for its efficiency on small and nearly sorted datasets. Intro Sort, as a combination algorithm that integrates Quick Sort, Heap Sort, and Insertion Sort, was studied to reveal how its adaptive mechanism can provide more optimal results. The testing was conducted by measuring execution speed, sorting stability, and memory usage efficiency. The findings from this study show that Bubble Sort ranks lowest in terms of performance and is less suitable for large data sets. Insertion Sort shows better results on small data sets and those with similar patterns. Intro Sort emerges as the most effective algorithm with stable processing time, high adaptability, and faster and more efficient sorting results for various data sizes. Overall, this study emphasizes the importance of choosing a sorting algorithm that suits the characteristics of the data and the needs of the application. The combination of adaptive strategies such as those in Intro Sort is the best solution for current data processing, which demands high speed and efficiency.

Comparison of Balancing Strategies for Classifying Guava Fruit Diseases

2025-11-27T15:22:43+07:00

The problem of class imbalance often poses an obstacle in deep learning-based image classification, especially in the domain of digital agriculture. The imbalance in data distribution makes it easier for models to recognize the majority class, while performance for the minority class declines. This study aims to analyze the effectiveness of three strategies for handling class imbalance: Weighted Loss Function, Oversampling, and a combination of Weighted Loss and Oversampling, in improving the performance of image classification of guava fruit diseases using a transfer learning-based MobileNetV2 architecture. The dataset consists of 3,784 images of three disease classes, namely Anthracnose, Fruit_Fly, and Healthy_guava, which show an imbalanced distribution. The research was conducted through the stages of Exploratory Data Analysis (EDA), pre-processing, augmentation, model training with four scenarios, and evaluation using Accuracy, Precision, Recall, F1-Score, and Macro Average F1-Score. The results showed that the Combination model (Oversampling and Weighted Loss) performed best on the minority class with an F1-score of 0.9630, the highest among all models. The Oversampling strategy produced the highest Macro F1-score of 0.9617, while Weighted Loss provided a significant improvement in classification sensitivity but was still below the combination model. Thus, it can be concluded that the combination strategy is the most effective approach in improving the sensitivity of the model to minority classes, while Oversampling excels in the overall performance stability of the model.

Predicting Student Academic Performance Based on Learning Habits Using XGBoost and SHAP

2025-11-27T15:23:51+07:00

This study developed a model for predicting student academic achievement based on learning habits using the XGBoost algorithm and SHAP interpretability techniques. The secondary dataset contains 1,000 entries and 16 variables (for example, hours of study per day, mental health, frequency of exercise, social media use, hours of sleep) pre-processed including cleaning, imputation, encoding, and normalization before being divided into train–test (80:20) and validated using 5-fold CV. Three models were tested: Linear Regression, Random Forest, and XGBoost. Evaluation using RMSE, MAE, and R² showed that XGBoost achieved RMSE = 0.335, MAE = 0.266, and R² = 0.882, while Linear Regression showed the best performance according to R² in certain configurations (R² = 0.888; RMSE = 0.326). SHAP analysis revealed that the most influential features were hours of study per day, mental health scores, exercise frequency, duration of social media use, and hours spent watching Netflix. The findings confirm that students' study habits and psychological conditions are the main determinants of academic achievement variation; the use of interpretable features strengthens the readability of the model for education stakeholders. Research recommendations include testing the model on longitudinal datasets, integrating socioeconomic factors, and implementing data privacy procedures before institutional-scale implementation.

Optimization of Classification of Tea Leaf Disease Images Using LBP–HOG and MobileNetV2

2025-11-27T16:11:29+07:00

This study was motivated by the need for an accurate and efficient system for detecting tea leaf diseases, given that the current method Manual identification has limitations in terms of consistency, speed, and It also depends on expert labor. To address these challenges, the study It developed a classification model for detecting diseases in tea leaves using a combination of features Local Binary Patterns (LBP) and Histogram of Oriented Gradients (HOG) integrated with the MobileNetV2 architecture. The research method includes the following stages: importing the dataset, data partitioning, exploratory data analysis (EDA), preprocessing, features, and training four model scenarios: baseline MobileNetV2, LBP-based model, HOG-based model, and hybrid LBP–HOG model. Evaluation is done with the metrics of accuracy, precision, recall, and F1-score. The results show that the baseline model achieved 91.67% accuracy, the LBP model achieved 60.67%, the HOG model achieved 68.67% accuracy, and the hybrid model achieved 66.67% accuracy. These findings indicate that MobileNetV2 is still the most optimal model, but the integration of texture features and gradients provides a deeper understanding of the characteristics of disease patterns. This study emphasizes the importance of exploring classic features to enriching visual representation in lightweight CNN models, as well as providing a contribution to the development of plant disease diagnosis systems that are efficient.

Segmentation of Coffee Purchasing Behavior Based on Transaction Time Using the K-Means Algorithm

2025-11-28T08:50:05+07:00

This studyaims to identify customer behavior patterns based on the time of purchaseof beverages at a coffee shop using the K-Means method.Transaction data includes purchase time, payment type, product name,time category, day, and month. The research stages include data cleaning, time attribute transformation, and numerical feature normalization. The optimal number of clustersis determined through testing k = 2–10 with four evaluation metrics,namely Inertia, Silhouette Score, Davies–Bouldin Index, and Calinski–HarabaszIndex. Based on the validation results, k = 3 was selected because it provided the best balancebetween compactness and cluster separation. The clustering results showedthree main customer groups based on transaction time trends:nighttime buyers with a peak around 10:27 p.m., afternoon to early evening buyerswith a centroid of 7:01 p.m., and morning to noon buyers with a centroid11:13. The frequency distribution indicates that the morning–afternoon buyer groupis the largest, while the early evening–night group is thesmallest. Visualization of scatter plots, boxplots, and time category graphsemphasizes the differences in characteristics between clusters. Overall,this study proves that K-Means is effective in mapping the temporal patternsof customer behavior. These findings can be used to develop time-based marketing strategies, operational arrangements, and product stock management,as well as form the basis for further analysis in the industry.

Comparison of Graph-Based Filtering and Non-Local Means Techniques in Diabetic Retinopathy Classification

2025-11-28T19:04:46+07:00

Classification of diabetic retinopathy (DR) based on retinal images is important for early detection, but is often hampered by poor image quality such as noise, uneven lighting, and low contrast. This study analyzes the effect of applying three image filtering techniques, namely Graph Laplacian Filtering (GLF), Graph Convolutional Network (GCN), and Non-Local Means (NLM), on improving the performance of Diabetic Retinopathy classification. The three methods were compared with a baseline model without filtering using VGG16 and evaluated through accuracy, AUC, loss, and image quality metrics such as PSNR, SSIM, MSE, and RMSE.The results showed that graphical and spatial filtering did not always improve classification performance, as VGG16 Fine-Tuning without filtering achieved the highest accuracy of 97.84%. Combinations with NLM, GCN, and Graph Laplacian resulted in lower accuracy due to the smoothing effect that removed important microfeatures on the retina. However, NLM remained effective in reducing noise without disturbing edge structures. These findings confirm that improving image visual quality does not always correlate with CNN accuracy, so preprocessing must focus on preserving diagnostic features.

IoT Application in Cashier Systems to Help People with Disabilities (Deaf)

2025-11-28T19:25:37+07:00

The rapid development of the Internet of Things (IoT) has brought significant impact across various aspects of human life, including information systems and public services. One of its important applications lies in supporting inclusivity for people with disabilities. This research focuses on the implementation of IoT in a cashier system specifically designed to assist individuals with hearing impairments in conducting payment transactions more easily, independently, and equally. In conventional cashier systems, most transaction information is delivered through audio signals, which creates a barrier for hearing-impaired users in fully understanding the payment process. To address this issue, this study develops and implements a prototype of an IoT-based cashier system that utilizes visual notifications and digital indicators as the main medium for delivering transaction information.The results of testing indicate that the IoT-based cashier system functions effectively in delivering transaction information, reducing communication errors, and improving the independence of hearing-impaired users during the payment process. Therefore, this research contributes not only to the development of modern cashier systems but also to the advancement of inclusive and accessible technology that supports equal opportunities for all members of society.

Sentiment Analysis of TikTok User Comments on The Free Nutritious Meal Program Using Support Vector Machine

2025-11-29T17:20:42+07:00

This study aims to analyze user sentiment when leaving comments on TikTok about the Free Nutritious Food Program (MBG) to understand how the public views the program. Comment data was obtained through online collection and then divided into three groups: positive, negative, and neutral. Before further processing, the data went through a text cleaning and stemming stage to reduce word variation. The data was then represented using the TF-IDF method before being classified with a Support Vector Machine algorithm. The evaluation results showed that using stemming provided more accurate results than without using stemming, thereby improving the model's ability to recognize sentiments contained in comments using informal language. Additional analysis using word clouds, n-grams, and topic modeling provided an overview of words and issues frequently appearing in public discussions regarding the program.

Sentiment Analysis of “Cek Bansos” Application Reviews on Google Play Store Using the Naïve Bayes Algorithm

2025-11-30T09:40:11+07:00

The rapid development of digital public services requires a deeper understanding of user perceptions and experiences regarding government applications, including Cek Bansos. This study aims to identify the polarity of user reviews by applying the Multinomial Naïve Bayes algorithm to review data collected from the Google Play Store. The methodology includes text preprocessing, sentiment labeling, feature extraction using TF–IDF, and model training and evaluation based on accuracy, precision, recall, and F1-score. The results show that the model achieves an accuracy of 79.5%, with very high performance in the negative class (recall 0.97) but poor performance in the neutral class due to data imbalance. The dominance of negative sentiment in the dataset indicates that users face significant technical difficulties, particularly in registration, verification, and service access. These findings demonstrate that Multinomial Naïve Bayes is effective as a baseline model for sentiment analysis; however, improving data balance and quality is necessary to produce a more stable, accurate, and representative model for evaluating digital public services.

Implementation of an Executive Information System for Thesis Document Submission with the Addition of AES-256-CBC Cryptography Algorithm

2025-10-31T15:45:20+07:00

The rapid digitalization of higher education demands secure and efficient management of academic documents such as thesis submissions. This study aims to develop an Executive Information System (EIS) for Thesis Document Submission integrated with AES-256-CBC cryptographic security to ensure data confidentiality, integrity, and controlled access. The system is implemented as a web-based platform using the Laravel framework and MySQL database, where each uploaded thesis document is automatically encrypted, and only authorized users with a valid Master Key can decrypt it. The AES-256-CBC algorithm generates unique ciphertexts for every encryption process, supported by randomized Initialization Vectors and separate key management to prevent unauthorized access or data leakage. Furthermore, the EIS dashboard implements the drill-down method, presenting real-time analytical information. This allows academic leaders to navigate hierarchically from high-level summaries to specific, detailed data, enhancing their ability to monitor thesis submissions and make informed decisions effectively. The results indicate that the integration of cryptography and executive information management enhances both document security and administrative efficiency, providing a reliable and transparent solution for safeguarding academic data within higher education institutions.

Investment Policy Analysis Between Generation Z and Millennial Generation

2025-11-20T21:12:25+07:00

This study aims to analyze and This study uses a quantitative approach with a comparative associative survey method and involves 100 respondents (50 Gen Z and 50 Millennials) who are domiciled in Indonesia and have investment experience, with a purposive sampling technique . comparing investment policies between Generation Z and Millennials in Indonesia. These two demographic groups are "Digital Natives" who are familiar with the development of information technology and investment applications. Primary data were collected through an online questionnaire and analyzed using Multiple Linear Regression and Independent Sample t-test with the help of the SPSS program.The results of the regression analysis show that financial literacy (X1), investment knowledge (X2), and investment risk (X3) simultaneously and partially have a significant effect on investment policy (Y) in both generations. The coefficient of determination (R2) value ^of0.652 indicates that 65.2% of the variation in investment policy can be explained by these three variables.The results of the comparative test ( Independent Sample t-test ) with a Sig. (2-tailed) value = 0.018 (<0.05) indicate a significant difference in investment policies between Generation Z and the Millennial Generation.Generation Z (born 1996–2010) tends to be more daring in taking risks and is quicker in adopting digital investment technologies such as applications (Bibit, Bareksa, Ajaib, IPOT), and chooses instruments with medium to high risk (stocks and digital assets).The Millennial generation (born 1981–1995) exhibits more conservative behavior , oriented towards the security and stability of assets , with a preference for relatively low-risk instruments such as mixed mutual funds, deposits, and blue-chip stocks.These findings reinforce Behavioral Finance theory , emphasizing the importance of individual understanding, experience, and risk perception in investment decision-making. The research's implications suggest the need for the government/OJK to expand digital financial literacy programs tailored to each generation, as well as the development of educational and secure features by securities firms.

Analysis and Visualization of Sales Transaction Patterns using Decision Tree and Tableau Public

2025-11-21T19:08:39+07:00

This study aims to analyze sales transaction patterns of rubber waste at PT Mandiri Enviro Technosio by integrating the Decision Tree algorithm with interactive visualization using Tableau Public. The dataset consists of 405 sales transactions recorded during the 2024–2025 period, comprising attributes such as transaction date, product type, quantity, unit price, total value, delivery region, and buyer category. The research methodology includes data acquisition, preprocessing to ensure data quality and consistency, construction of a classification model using the CART algorithm, evaluation of model performance through a confusion matrix, and development of interactive dashboards for enhanced interpretability. The Decision Tree model achieved an accuracy of 88.24% in classifying transaction values into low, medium, and high categories. Unit price and transaction period were identified as the most influential attributes in determining transaction value. Visualization using Tableau Public effectively presented the distribution of transaction values, sales trends, and geographical patterns, thereby strengthening analytical insights and supporting data-driven decision making. The integration of classification techniques and interactive visualization contributes to improving business intelligence capabilities and enables the formulation of more adaptive, evidence-based sales strategies.

Mitigating Imbalanced Citrus Disease Image Datasets with Oversampling

2025-11-28T08:50:46+07:00

Dataset imbalance is a critical challenge in plant disease image classification because it causes bias towards the majority class. This study evaluates the effectiveness of augmentation-based oversampling techniques on the classification performance of citrus leaf images using the MobileNetV2 architecture. The four leaf disease classes classified include Greening, Fresh, Canker, and Blackspot. The dataset was obtained from a public repository and processed through preprocessing (resize, normalization) and augmentation (rotation, flipping, zoom) stages. The model was trained and tested in two scenarios: baseline (unbalanced data) and mitigation (data balanced through augmentation). The experimental results show that the mitigation approach was able to increase accuracy from 91.92% to 93.94%. The F1-score, precision, and recall values also increased significantly, especially in the minority class. Evaluation using a confusion matrix reinforced the finding that augmentation-based oversampling is effective in reducing classification errors. This study shows that the integration of augmentation techniques and MobileNetV2-based transfer learning can significantly improve classification performance and contribute to the development of early detection systems for plant diseases in precision agriculture.

Comparative Analysis of Durian Leaf Disease Classification Using Transfer Learning VGG16, InceptionV3, and U-Net

2025-11-28T08:49:12+07:00

Image-based durian leaf disease detection presents challenges due to high visual similarity among symptoms and the limited, imbalanced dataset. This study compares three deep learning architectures VGG16, InceptionV3, and U-Net encoder-based—using transfer learning for classifying five durian leaf conditions. The dataset of 4,437 images underwent preprocessing, augmentation, and preliminary segmentation using U-Net to enhance focus on leaf regions. Fine-tuning was applied to the upper layers of each model to adapt feature representations to tropical leaf characteristics. The results indicate that InceptionV3 achieved the most stable and accurate performance with an accuracy of approximately 0.66, while VGG16 showed balanced results but was more prone to overfitting. U-Net proved effective for segmentation but less optimal as a classifier due to loss of small-scale lesion details. Overall, the findings demonstrate that combining U-Net segmentation with CNN-based transfer learning improves disease identification performance, particularly under limited data conditions.

Optimizing the Execution Time of JOIN Queries and Subqueries Using MySQL

2025-11-28T19:08:54+07:00

Relational database systems form the backbone of modern information management. However, the escalating volumes of data and increasing complexity of queries present substantial performance challenges in data retrieval operations. This study investigates the execution time differences between Subqueries and five join methods: Inner Join, Left Join, Right Join, AsOf Join and Lateral Join, in MySQL environments. An experimental methodology was employed, utilising two simulated relational tables containing 100, 1,000, and 10,000 rows of data. Each query method was executed three times under identical system conditions to establish reliable average execution times. The findings demonstrate that join operations substantially outperform subqueries across all tested datasets. Inner Join, Left Join and Right Join maintained execution times below 0.04 seconds, even with the most extensive dataset. Conversely, subqueries exhibited significant performance degradation, with execution times increasing to tens of seconds as the data volume increased. This performance disparity stems from the iterative processing inherent to subqueries, which intensifies proportionally with dataset scale, whereas join operations leverage more efficient simultaneous data processing and merging algorithms. The research concludes that join methods constitute the more appropriate choice for medium to large-scale data scenarios, offering practical optimisation guidance for database developers and administrators implementing MySQL-based systems.

Optimization of Convolutional Neural Networks Using Resizing Techniques for Banana Leaf Disease Classification

2025-11-29T18:43:35+07:00

Early and accurate identification of banana leaf diseases is essential for supporting digital agriculture, as visual symptoms often require rapid and reliable analysis. This study investigates the impact of three image resizing techniques squashing, letterboxing, and random resized crop on the performance of the MobileNetV2 architecture in classifying four categories of banana leaf images using the Banana Leaf Disease Dataset v4 consisting of 4,675 samples. The experiments were conducted using a transfer learning approach with an 80:10:10 data split, standardized normalization, and data augmentation. The results show that all resizing techniques achieved test accuracies above 92%. Squashing produced the highest accuracy and fastest training time, letterboxing demonstrated the most stable performance with the lowest validation loss, and random resized crop improved generalization to variations in object position. These findings confirm that resizing strategies significantly influence the stability and effectiveness of CNN models. Overall, MobileNetV2 proves capable of delivering accurate and efficient classification of banana leaf diseases when supported by an appropriate preprocessing pipeline. This study provides empirical evidence for developing image-based plant disease diagnosis systems within smart agriculture.

Comparison of Logistic Regression and XGBoost Model Performance in Predicting Credit Scores

2025-11-29T17:23:20+07:00

Credit Scoring is a mathematical approach used to assess the creditworthiness of individuals or companies by classifying debtors into certain categories based on their risk profiles. This study aims to compare the performance of the Logistic Regression and XGBoost machine learning algorithms in predicting credit scores (credit scoring) to reduce the risk of Non-Performing Loan (NPL) risk at PT Graha Mazindo Mandiri. The secondary dataset used contains 1,533 car loan debtor data with 17 variables, including 1dependent variable and 16 independent variables. The research process includes data preprocessing (cleaning, handling outliers, encoding, normalization, and class balancing with SMOTE), modeling, and evaluation using the Accuracy, Precision, Recall, F1-score, and ROC-AUC metrics. The results show that XGBoost excels with 96% accuracy and ROC-AUC of 0.99 compared to Logistic Regression with an accuracy of 88% and ROC-AUC0.94, due to XGBoost ability to capture non-linear patterns and handle data imbalance. This study provides insights into credit risk factors and supports more accurate credit decision-making, with recommendations for hyperparameter optimization and model integration into operational systems.

Automatic Bell Using Esp8266 and Telegram Method as a Reminder for Laboratory Time at the AMIKOM Purwokerto University Assistant Forum

2025-11-29T17:19:54+07:00

The purpose of this research is to create an automatic bell system that uses an ESP8266 microcontroller integrated with Telegram as a reminder for practical sessions at the Amikom Purwokerto University Assistant Forum. This system is necessary because assistants need to balance laboratory responsibilities and academic activities. Using an Internet of Things-based approach, this system combines NodeMCU ESP8266, DS3231 Real-Time Clock (RTC) module, buzzer, and Telegram Bot notification service. The research process includes identifying needs, reviewing literature, designing the system, implementing, and testing. The bell operates automatically according to the schedule stored in the RTC, while the Telegram bot sends reminders 15 minutes before the practicum begins. Test results show that the bell consistently activates at the right time without delay, and that Telegram notifications are sent according to the configured schedule. These results indicate that the proposed system can meet the functional requirements for accuracy, reliability, and effective communication. Potential for further development in this system includes integration with an automatic attendance feature.

Application of Weighted Loss Function in Convolutional Neural Network for Acne Image Classification

2025-11-30T14:48:11+07:00

Automated acne image classification using Convolutional Neural Networks (CNN) holds significant potential in dermatological diagnosis but faces a fundamental challenge of class imbalance. This phenomenon causes standard models to be biased towards majority classes and fail to recognize clinically important minority classes. This study aims to address this bias by applying a Weighted Loss Function to the EfficientNetB1 architecture. The research method employs a comparative experimental approach between two scenarios: the Baseline model (Standard Cross-Entropy) and the Proposed model (Weighted Cross-Entropy). The dataset consists of 5 acne classes with an imbalanced distribution. The results show that the Weighted Loss model significantly outperforms the Baseline model. Overall accuracy increased from 80% to 86%. The most significant improvement occurred in the minority class 'Papules', where the F1-Score surged by 0.10 points (from 0.71 to 0.81). It is concluded that the application of Weighted Loss Function effectively overcomes bias due to imbalanced data without the need for synthetic data augmentation, resulting in a fairer and more reliable model for clinical implementation.

Influence of AI Technology on the Development of Critical Thinking Skills in Education

2025-12-01T11:31:45+07:00

The development of Artificial Intelligence (AI) technology has brought significant changes to the educational process, especially in supporting the development of critical thinking skills in students. This study uses a qualitative approach with a systematic literature review method to analyze various findings regarding the influence of AI in education. The results of the study show that AI has the potential to improve critical thinking skills through the provision of analytical stimuli, adaptive feedback, and the facilitation of active learning that encourages reflection, evaluation, and data-based argumentation. However, excessive use of AI or use without teacher guidance can lead to dependence, reduce creativity, and weaken students' evaluative and independent thinking skills. Therefore, the integration of AI should be balanced with active learning strategies, digital literacy, and pedagogical guidance so that this technology functions as a cognitive partner that strengthens critical thinking processes, rather than as a substitute for students' reasoning.

Prediction of Clean Water Quality Using K-Nearest Neighbor (KNN) and Naïve Bayes at PDAM Kupang City

2025-12-01T11:26:28+07:00

Kupang City faces significant challenges in providing clean water due to its dry geographical conditions and extreme climate. Although it has various potential water sources such as watersheds and bore wells, clean water distribution remains suboptimal. This study aims to predict clean water quality using two machine learning algorithms, namely K-Nearest Neighbor (KNN) and Naïve Bayes, based on the Water Quality Dataset which includes parameters such as pH, hardness, total dissolved solids, and turbidity. The process involves data preprocessing, algorithm implementation, and model evaluation using classification metrics. The KNN model achieved an accuracy of 56%, with an F1-score of 0.67 for the “unsafe” class and 0.36 for the “safe” class. Meanwhile, the Naïve Bayes model achieved a higher overall accuracy of 61% but failed to detect the “safe” class, showing a precision and recall of 0.00. Overall, KNN performed more balanced across classes despite its moderate accuracy, while Naïve Bayes was biased toward the majority class. These findings highlight the importance of selecting appropriate algorithms and tuning parameters for water quality prediction. The implementation of predictive models is expected to assist PDAM Kupang in making data-driven decisions to improve clean water management sustainably.

Sentiment Analysis of Public Opinion on RUU KUHAP 2025 Using Multinomial Naïve Bayes and Random Oversampling

2025-12-02T09:19:54+07:00

The ratification of the Draft Criminal Procedure Code (RUU KUHAP) in 2025 has triggered a significant wave of public reaction on social media, particularly on YouTube. Understanding these public sentiments is crucial for evaluating the legislative performance of the House of Representatives (DPR). This study aims to classify public opinion into positive and negative sentiments using the Multinomial Naïve Bayes algorithm. The dataset consists of 2,370 user comments collected from YouTube. To address the chal lenge of unstructured text, a comprehensive pre - processing pipeline was implemented, including cleaning, normalization, and stemming. Furthermore, this research addresses the issue of class imbalance , where negative comments dominated (73.9%) by applying the Random Oversampling (ROS) technique to the training data. The feature extraction was performed using TF - IDF. The experimental results demonstrate that the proposed model achieved an overall Accuracy of 87.22%. Detailed evaluation shows a Pr ecision of 0.9 1 and Recall of 0.93 for the negative class, confirming the model's robustness. These findings indicate that the majority of public sentiment is critical of the RUU KUHAP , focusing on issues of corruption and trust. This research contributes to the field of text mining by demonstrating the effectiveness of oversampling in improving Naïve Bayes performance on imbalanced social media data.

Measurement of Digital Service Quality and Success of District Library Information System (SIPERKA) Implementation: Integration of HOT-Fit Model and Service Quality Dimensions

2025-12-02T15:21:32+07:00

This study evaluates the success of District Library Information System (SIPERKA) implementation in Subang Regency using the integration of HOT-Fit (Human, Organization, Technology-Fit) model and service quality dimensions. Employing SEM-PLS methodology with 115 village librarians and library operators, the research examines how Technology components (system quality, information quality, service quality), Human factors (user satisfaction, user competency, system use), and Organizational aspects (organizational structure, leadership support, environment) collectively influence net benefits. Results demonstrate that service quality exerts the strongest influence (β=0.312) on user satisfaction among technological dimensions, while system use (β=0.468) emerges as the primary determinant of net benefits. The integrated model explains 71.5%-76.8% variance in endogenous variables with Goodness of Fit (GoF) of 0.719, indicating excellent model performance. All ten hypotheses received empirical support (p<0.05). This research contributes theoretically by demonstrating the critical importance of service marketing perspective in public sector information systems evaluation, revealing that service quality supersedes technical quality in determining user satisfaction. Practically, it provides evidence-based recommendations for improving digital service quality in village libraries, with documented Return on Investment (ROI) of 630.8% demonstrating SIPERKA's success in elevating village library data achievement from below 40% to 87%.

Sentiment Analysis of Honda Esaf Frame Quality Based on Reviews on Platform X using Support Vector Machine Algorithm

2025-12-02T15:23:24+07:00

This study analyzes public sentiment towards Honda's eSAF frame through 1513 reviews on Platform X during the period of January 2023-October 2025, which was triggered by crucial issues related to the potential for rust, corrosion, and fracture in motorcycle frames. Using a quantitative method with a computational approach, this study applies the Support Vector Machine (SVM) Algorithm with data preprocessing (Case Folding, Cleaning, Tokenizing, Stopword Removal, Stemming), TF-IDF weighting, and Lexicon-based sentiment labeling to classify positive and negative perceptions. The evaluation results show that the SVM-TF-IDF model achieved 98% accuracy on the test data, with negative sentiment dominated by the keywords "rust" and "damaged", while positive sentiment centered on "strong" and "safe", providing an objective picture of public perception as a basis for evaluating product quality and improving corporate communication strategies.

The Relationship Between Hedonism Lifestyle and Student Consumer Behavior in Pamekasan District

2025-12-02T19:09:13+07:00

Technological developments and globalization have encouraged the emergence of a hedonistic lifestyle among college students, characterized by a tendency to pursue pleasure, luxury, and trend-driven consumption. This situation has the potential to influence student consumer behavior, including excessive purchasing and a lack of consideration for real needs. This study aims to analyze the relationship between a hedonistic lifestyle and student consumer behavior in Pamekasan Regency. The study used a quantitative approach with a correlational approach.

The sample consisted of 104 students selected through purposive sampling. Data collection was conducted through an online questionnaire with a Likert scale. Validity tests using Pearson Product Moment correlation showed all items were valid, while reliability tests yielded a Cronbach's Alpha value of 0.956, indicating high reliability of the instrument. Normality tests showed the data were normally distributed (sig. X = 0.080; Y = 0.070). Spearman's Rho correlation test yielded a coefficient value of 0.780 with a significance level of 0.000.

The results of this study indicate a strong, positive, and significant relationship between a hedonistic lifestyle and student consumer behavior. This means that the higher the level of hedonism in students, the higher their tendency to engage in consumer behavior. Therefore, a hedonistic lifestyle is a significant factor influencing student consumption patterns in Pamekasan Regency.

This study concluded that the higher the level of hedonism in students, the higher their tendency to engage in consumer behavior. These findings are expected to serve as a guide for universities, parents, and students in understanding and managing consumption patterns to be more rational and based on priority needs.

Sentiment Analysis of Mie Gacoan Pemuda Cirebon Restaurant Reviews Using Support Vector Machine

2025-12-02T20:38:09+07:00

The growth of digital platforms has increased the use of sentiment analysis to understand public perceptions of business services. Customer reviews on Google Maps provide valuable insights but are unstructured and linguistically diverse, requiring robust analytical methods. This study conducts sentiment analysis on reviews of Mie Gacoan Pemuda Cirebon using a Support Vector Machine (SVM) classifier. The research focuses on designing an effective text preprocessing pipeline, identifying sentiment distribution, and evaluating SVM performance. The methodology includes web scraping, manual labeling, text preprocessing, TF-IDF feature extraction, dataset splitting, model training, and evaluation using accuracy, precision, recall, and F1-score. The results show that the majority of reviews are positive, and the SVM model achieves strong performance with an accuracy of 0.82. These findings provide an objective overview of customer perceptions and demonstrate the effectiveness of SVM for Indonesian-language sentiment classification. The model can support businesses in improving service quality based on customer feedback.

Classification of Pneumonia Using CNN and Vision Transformer

2025-12-03T11:59:40+07:00

Pneumonia remains one of the leading causes of mortality among children worldwide. This study aims to evaluate the performance of two deep learning architectures, Convolutional Neural Network (CNN) and Vision Transformer (ViT), for pneumonia classification using chest X-ray images. Four training scenarios were examined, consisting of MobileNetV2 baseline, MobileNetV2 fine-tuned, ViT baseline, and ViT fine-tuned models. The dataset was obtained from the Chest X-Ray Images (Pneumonia) collection and was processed through augmentation and preprocessing to produce a balanced set of 9,000 images. Baseline models were trained using a feature extraction approach, while fine-tuning was conducted by selectively unfreezing internal layers. Experimental results show that all models achieved accuracy above 95%. The MobileNetV2 baseline reached 97.63%, while its fine-tuned counterpart did not yield further improvement, achieving 97.41%. In contrast, the Vision Transformer demonstrated substantial performance gains, where partial fine-tuning produced the highest accuracy of 98.59% with an f1-score of 0.99. These findings indicate that ViT with targeted fine-tuning is more effective in capturing global representations within X-ray images, making it a strong candidate for computer-aided pneumonia detection systems supported by artificial intelligence.

Basic Analysis of Cybersecurity in Facing Digital Threats in the Industrial Era 5.0

2025-12-03T15:28:16+07:00

Industri 5.0 fokus pada kerja sama yang lebih mengutamakan manusia, serta keberlanjutan dan ketahanan. Teknologi canggih seperti kecerdesan buatan (IA), internet of things industri (IIOT), dan robot yang bekerja bersama manusia (cobot) terintegrasi ke dalam industri ini. Karena adanya keterhubungan yang lebih baik antara dunia fisik dan dunia digital, maka peningkatan ini berdampak lebih besar pada peningkatan risiko serangan digital, sehingga mengancam data, sistem, dan bahkan keselamatan manusia. Penelitian ini bertujuan untuk menganalisis secara mendalam fondasi keamanan siber dalam konteks industri 5.0, serta menemukan strategi yang bisa diterapkan dalam menghadapi ancaman digital yang terus berkembang. Metode yang digunakan adalah observasi bahan bacaan secara sistematis dan analisis deskriptif kualitatif terhadap kerangka kerja keamanan siber yang ada, seperti NIST, ISO 27001, dan IEC 62443, terutama dalam konteks teknologi industri 5.0. hasil penelitian menunjukkan bahwa perlu ada perubahan dari perlindungan yang hanya di sekitar batas fisik ke model keamanan yang lebih proaktif, terdistribusi, dan berdasarkan risiko. Model ini menekankan pentingnya arsitektur zero trust, Perlindungan data yang menyeluruh, dan pemantauan ancaman yang ditingkatkan oleh IA. Selain itu, kesadaran dan pelatihan tenaga kerja juga ditemukan sebagai bagian penting dari kemanan siber.

The Effect of E-Wallet Usage on Personal Cash Flow and Net Worth Ratio in Generation Z

2025-12-07T17:35:36+07:00

The rapid development of financial technology has significantly transformed individual digital financial behavior, particularly through the increasing use of electronic wallets (e-wallets) among Generation Z. As digital natives, this generation is highly exposed to online transactions, yet their financial management capabilities remain varied. This study aims to analyze the effect of e-wallet usage on personal financial stability, specifically measured through personal cash flow and net worth ratio. Additionally, technological adaptation patterns within modern student financial activities significantly increase complexity, influencing how digital tools are utilized. A quantitative survey method was employed, involving 30 respondents who are active e-wallet users and university students in Pamekasan, Madura. Data were collected through a structured questionnaire and tested for reliability, yielding a Cronbach’s Alpha value of 0.761, indicating acceptable internal consistency.

The Kolmogorov–Smirnov normality test showed that some variables met the normal distribution criteria. Results of multiple linear regression revealed that the intensity of e-wallet use, perceived usefulness, and perceived financial impact did not have a significant effect on cash flow, with a significance value greater than 0.05. The model’s R Square value of 0.087 further suggests that only 8.7% of changes in cash flow can be explained by the examined variables, while the remaining 91.3% is influenced by factors such as income level, spending behavior, and financial literacy. These findings indicate that although e-wallets have become an integral part of students’ daily transactions, their impact on overall financial stability remains limited. Strengthening digital financial literacy is recommended to promote wiser and more responsible e-wallet usage.

Analysis of the Effectiveness of Manual Deployment and CI/CD Github Actions in the Braisee Application

2025-12-05T21:27:29+07:00

In the modern cloud-based software development ecosystem, the speed and reliability of the deployment process are critical elements. This study aims to evaluate the effectiveness of implementing Continuous Integration/Continuous Deployment (CI/CD) using GitHub Actions compared to manual methods for the machine learning API of the Braisee application hosted on Google Cloud Run. Using a quantitative approach with a comparative experimental design across ten testing iterations, this research measures deployment time efficiency, error rates, and system stability. The experimental results show a significant performance disparity, where the automated method based on GitHub Actions is considerably more efficient, with an average total duration of 111–167 seconds, reducing operational time by 40–60% compared to the manual method, which requires 297–364 seconds. In terms of reliability, the automated method achieves a 100% success rate with high consistency, whereas the manual method demonstrates substantial vulnerability to human errors such as mistyped project IDs and inconsistent image tagging. It is concluded that implementing CI/CD through GitHub Actions is a superior solution that improves time efficiency and ensures the stability of cloud-based applications compared to manual procedures.

Application of Machine Learning in Predicting FIFA World Cup Matches

2025-12-07T07:12:36+07:00

Football is one of the world’s most widely followed sports, making it an appealing subject for predictive analytics using modern data technologies. This study aims to build a predictive model for international football match outcomes by applying the CRISP-DM methodology as the analytical framework. The dataset used is international_matches.csv covering the period 1993–2022, which underwent a series of preprocessing steps including data cleaning, feature engineering, encoding, imputation, and scaling. Several machine learning algorithms were evaluated, namely Logistic Regression, Random Forest, and HistGradientBoostingClassifier (HistGBM). The best model was obtained using the optimized HistGBM, which demonstrated superior capability in identifying home-team victories, achieving a Recall of 78%. This high sensitivity indicates that comparative features—such as rank difference and squad strength disparity across goalkeeper, defense, midfield, and attack attributes—play a crucial role in predicting dominant match outcomes. The trained model was subsequently deployed into an interactive Streamlit-based web application that enables users to input match-related information and obtain real-time predictions. Overall, this study shows that machine learning methods can be effectively utilized to support data-driven analysis of international football match outcomes.

Comparative Analysis of Serverless Container Service Performance Between Google Cloud Run and AWS App Runner in Cross-Cloud Architecture

2025-12-07T07:14:17+07:00

Research on the performance of serverless container services is becoming increasingly important as the need for modern distributed and cross-cloud architectures grows. This study analyzes the performance of two leading serverless services, Google Cloud Run and AWS App Runner, in a cross-cloud architecture scenario. Testing was conducted using identical parameters, including container configuration, region, memory, vCPU, and concurrency. Performance testing included p95 latency, throughput, and error rate metrics using loads of up to 1000 virtual users. The results showed that Google Cloud Run provided more stable performance with p95 latency of 47–71 ms, throughput of 436–438 RPS, and 0% error rate. In contrast, AWS App Runner showed p95 latency of 490–651 ms with throughput variation of 388–410 RPS and an error rate of 2–4.41%. The difference in performance was due to autoscaling mechanisms, cross-cloud communication overhead, and resource contention. This study provides empirical evidence for selecting the optimal serverless service for distributed architectures.

Visualization of Lecturer Teaching Evaluation Data Using K-Means Clustering and Tableau Methods

2025-12-07T17:33:57+07:00

In the process, the results of monitoring and evaluating lecturers in each semester are usually only presented in the form of tables and descriptive explanations, but have not yet visualized the data for further analysis. The purpose of this study is to visualize the results of lecturer teaching evaluation using the K-Means Clustering and Tableau algorithms, and is expected to help the faculty and university monitor and evaluate lecturers in each semester in a more objective and informative manner. The results of the study found that the k-means clustering algorithm succeeded in finding the pattern of student clustering on the evaluation of lecturer teaching and based on the visualization of the results of k-means with a tableau it was found that most students gave a positive response to lecturer teaching and only a small number of students gave a poor assessment of lecturer teaching by emphasizing on improving the teaching process, namely consistently carrying out RPS, punctuality and so on

E-Commerce Customer Segmentation Application Based on the K-Means Algorithm

2025-12-09T08:33:27+07:00

Ineffective e-commerce marketing serves as the background for this research, which aims to develop a customer segmentation application for targeted marketing. The K-Means Clustering method with RFM (Recency, Frequency, Monetary) analysis is applied to data from 178 customers. The research methodology includes data preprocessing, feature transformation, and the determination of the optimal K using the Elbow Method. The results indicate that K=3 is the optimal number of clusters. Three segments were successfully identified: 'Champions' (18.5%, 33 customers) with the highest Frequency/Monetary values, 'Active & Potential' (41%, 73 customers) with the lowest Recency (most recent), and 'At Risk' (40.5%, 72 customers) with the highest Recency (longest duration since last transaction). The study concludes that the developed Streamlit-based application successfully visualizes these segments interactively to support strategic decision-making in marketing.

Implementation of the C4.5 Decision Tree Algorithm to Determine Student Productivity Based on Sleep Patterns

2025-12-09T13:19:09+07:00

Sleep patterns refer to an individual’s habits in managing sleep and wake times, including duration, quality, and regularity. Students, particularly those in the Informatics Engineering Program at Universitas Harapan Medan, often experience irregular sleep patterns due to heavy academic workloads such as assignments, projects, and practical activities. This condition can reduce academic productivity in terms of concentration, memory, and the ability to complete tasks on time. Therefore, this study aims to develop a classification model to predict student productivity levels based on sleep patterns using the Decision Tree C4.5 algorithm. This algorithm was chosen for its advantages in interpretability, ability to handle both numerical and categorical data, and efficient attribute selection, which contribute to generating an accurate and transparent classification model. The study involved 30 respondents from the 8th semester of the Informatics Engineering Program at Universitas Harapan Medan in the 2024/2025 academic year who filled out questionnaires regarding their sleep patterns and productivity. The results showed that 15 respondents (41.2%) had low productivity, 9 respondents (35.3%) had medium productivity, and 6 respondents (23.5%) had high productivity. These findings indicate a significant relationship between sleep pattern regularity and student productivity levels. The model generated using the C4.5 algorithm is expected to serve as a foundation for developing decision support systems aimed at improving the balance between sleep patterns and academic productivity among students.

Website-Based School Financial Information System

2025-12-10T14:07:19+07:00

The rapid development of information technology has brought significant changes in data and information management in the educational environment. This research focuses on the development of a website-based Financial Information System tailored to the needs of SDIT Binaul Izzah Bumiayu. This system was designed using the waterfall method with stages of needs analysis, design, implementation, and testing to improve efficiency, accuracy, and ease in managing school financial data which was previously still manual. The results of the study indicate that the system created is able to assist schools in the process of recording, recapitulating financial reports, and accelerating data access while reducing human error. Recommendations for future system development include the addition of payment notification features, automatic reports, student data integration, improvement of supporting facilities, and ongoing cooperation between schools and universities, followed by periodic evaluations to improve system quality.

Comparison of LSTM and ARIMA Methods in Predicting the Inflation Rate in Manado City

2025-12-12T11:00:25+07:00

Forecasting city-level inflation is challenging due to seasonal patterns, nonlinear dynamics, and limited exogenous variables, while short-term accuracy is required for timely policy responses. This study focuses on monthly inflation in Manado City over the period 2010–2024, explicitly accounting for the role of the Consumer Price Index (CPI). We compare a seasonal SARIMA baseline with a multivariate LSTM model that jointly ingests inflation and CPI series. The contributions of this work are an end-to-end, reproducible forecasting pipeline and an evidence-based comparison that identifies the conditions under which a feature-rich nonlinear model is preferable. The methodology includes aligning and preprocessing monthly series, conducting stationarity tests, selecting SARIMA specifications via information criteria and residual diagnostics, and training a 12-month window LSTM (Adam optimizer, MSE loss) with internal validation. The results show that the LSTM yields lower errors on the test horizon (RMSE 0.497; MAE 0.398) than the SARIMA (1,1,1)×(1,1,1,12) model (RMSE 0.661; MAE 0.486), with a smoother 12-month-ahead forecast path under a constant-CPI scenario; visual findings are consistent with the metrics, and a Diebold–Mariano test can be used to assess the significance of the difference. In conclusion, although SARIMA remains a strong and interpretable baseline, the multivariate LSTM delivers a practically meaningful gain in short-term accuracy when the inflation–CPI interaction is nonlinear, making it relevant for regional policy planning.

People Counting in Sample Video Footage Using CNN Integrated with YOLOv5

2025-12-11T21:13:52+07:00

Accurate people counting in dynamic environments remains challenging due to variations in lighting, complex backgrounds, and occlusion. This study proposes a video-based people counting system leveraging a Convolutional Neural Network (CNN) integrated with the YOLOv5 object detection model. The system applies a structured preprocessing pipeline, including frame extraction, normalization, and noise reduction, to enhance data consistency before detection. The model was evaluated using ten real-world campus video sequences to assess detection reliability and counting accuracy. Experimental results demonstrate that the proposed method achieves high precision and recall for real-time detection across diverse scenarios. Performance degradation was observed in frames containing dense crowds or low illumination, indicating limitations under extreme conditions. These findings validate the feasibility of lightweight CNN-based detectors for surveillance and monitoring applications, while highlighting the need for larger datasets and optimized training strategies to improve robustness in more complex environments.

Development of a Complaint Application for the Education Agency Using the Agile Development Method

2025-12-12T08:33:09+07:00

The rapid advancement of information technology has driven changes in performance and problem solving in society and government agencies. In the education sector, the Education Office, as the main service provider, requires an effective complaint mechanism for students, teachers, employees, and the community. The current complaint method still often relies on face-to-face submission, which causes problems such as difficulty in tracking the status of complaints, poor documentation, and limitations in evaluating the quality of complaint handling. This study aims to develop a web-based complaint application to digitize and harmonize the complaint process at the Minahasa Regency Education Office. The Agile Development method was used due to its iterative, flexible, and collaborative nature, allowing for the gradual development of features based on direct feedback from stakeholders. Data collection techniques included observation, interviews, documentation studies, and literature studies. The system was designed using UML diagrams, including Use Case, Activity, Sequence, and Class diagrams. Development was carried out in sprints, focusing on core features: user registration with NIK verification, complaint submission and tracking, and an admin dashboard for complaint management. Functional testing using the Black-Box method confirmed that all key features operate correctly as required. The resulting application successfully transformed the manual complaint process into a more structured, transparent, and efficient digital system, thereby contributing to improved public service quality in the field of education.

Implementation of IoT and Machine Learning for Monitoring and Prediction of Tank Water Levels

2025-12-12T08:31:29+07:00

The availability and quality of clean water in household storage tanks are essential yet often overlooked until problems such as depletion or contamination occur. Manual monitoring methods that rely on physical inspection tend to be inefficient, prone to delay, and unable to support predictive decision-making. This study proposes an automated monitoring solution by integrating Internet of Things (IoT) technology with Machine Learning-based analysis. The system is developed using an ESP32 microcontroller that continuously collects real-time data from an ultrasonic sensor to measure water level and a turbidity sensor to assess water clarity. The time-series data obtained is then analyzed using two algorithmic approaches. Linear Regression is employed to model the water depletion rate and generate predictions regarding the estimated remaining duration before the tank reaches an empty state. In parallel, Random Forest is applied as a comparative model to validate prediction accuracy under non-linear consumption patterns. Experimental results demonstrate that the combined IoT–Machine Learning framework provides accurate, timely, and informative insights for users. The proposed system improves water usage efficiency and strengthens early warning capabilities, making it a practical solution for supporting effective household water management.

Classification of Program Keluarga Harapan Assistance Recipients Using a Website-Based Support Vector Machine Algorithm (Case Study: Panyabungan Kota Subdistrict)

2025-12-12T13:15:36+07:00

Program Keluarga Harapan (PKH) is a social assistance program aimed at reducing poverty by providing financial aid to eligible families. This research focuses on the development and implementation of the Support Vector Machine (SVM) algorithm to classify PKH recipients in Panyabungan Kota Subdistrict, Mandailing Natal Destrict. The classification process utilizes factors such as family income, number of family members, and the presence of elderly members. These three factors are chosen due to their availability from public records, ensuring the privacy of participants. The classification model developed in this study is implemented in a web-based system built with PHP and JavaScript, designed to facilitate the automatic classification of PKH recipients. This system helps streamline the registration to be more precise and effective, providing an efficient solution for local government officials to identify eligible families for the PKH program. The evaluation results show that this system can classify PKH recipients well with an accuracy of 93%, offering an automated approach that supports decision-making in the distribution of social assistance.

Comparison of Random Forest and K-Nearest Neighbors in Heart Disease Prediction

2025-12-12T16:19:22+07:00

Heart disease is one of the leading causes of death worldwide, with a death toll reaching 17.9 million cases annually according to the World Health Organization (WHO) and a prevalence of 1.5% in Indonesia. This high mortality rate demonstrates the importance of early detection and accurate prediction to prevent more serious complications. The development of artificial intelligence technology, particularly machine learning, offers a new approach in the medical field through the ability to analyze clinical data quickly and efficiently. This study was conducted to compare the performance of two machine learning algorithms, namely Random Forest and K-Nearest Neighbors (KNN), in predicting heart disease using a clinical dataset from Kaggle containing 20 samples and 9 attributes related to the patient's physiological condition. The parameter optimization process in both algorithms was carried out using grid search techniques with cross-validation to obtain the best model that can perform optimally on a limited dataset. Performance evaluation was carried out using accuracy, recall, and precision metrics to comprehensively measure the quality of the model predictions. The results of the study showed that the Random Forest algorithm provided superior performance with an accuracy of 0.75, a recall of 0.88, and a precision of 0.86, compared to KNN which only achieved an accuracy of 0.50, a recall of 0.67, and a precision of 0.67. These findings indicate that Random Forest is more effective in identifying the presence of heart disease, especially in terms of sensitivity to positive cases and prediction consistency. Thus, Random Forest has the potential to be a more appropriate algorithm for implementation in machine learning-based clinical decision support systems, to support the process of diagnosing heart disease more accurately and efficiently.

Implementation of Prototyping Method in Developing a Web-Based Cos Management System using Laravel

2025-12-15T11:10:21+07:00

This study develops a web-based boarding house management system using the Prototyping method to address administrative issues at Kos D'Rosse, where data was previously managed manually through Google Form, Excel, and WhatsApp. The Prototyping model enabled iterative requirements gathering and user evaluation to refine system features. The system was built using the Laravel framework with an MVC architecture and includes modules for tenant management, room monitoring, payment processing, financial reporting, and a real- time dashboard. Blackbox testing confirmed that all features functioned according to user needs, while whitebox testing produced low Cyclomatic Complexity values, indicating simple and maintainable program logic. User Acceptance Testing (UAT) showed improvements in operational efficiency, data accuracy, and decision-making speed. The results demonstrate that the system integrates all management activities into a single platform, reduces administrative workload, and provides accurate, real-time information. Overall, the Prototyping approach and Laravel MVC support structured development and effective system performance.

Keywords: Laravel, Prototyping, Web-based System, Boarding House Management, MVC

Smart Safety Room: ESP32 Decision Tree-Based Multi-Hazard Detection System

2025-12-15T12:32:05+07:00

Physical space security and safety remain fundamental challenges in various sectors, ranging from residential buildings to critical server rooms. Conventional security systems often rely on single sensors or passive alarms that cannot respond comprehensively to multiple simultaneous threats. This research proposes a Smart Safety Room, an ESP32-based integrated multi-sensor security system that combines gas sensors (MQ-2), fire sensors (flame sensors), PIR sensors, and visual-audio output components including OLED displays, RGB LEDs, and buzzers. The system implements a decision tree algorithm with hierarchical priorities to classify room conditions into three categories: SAFE, ALERT, and DANGER based on a combination of sensor data. Testing was conducted through four main scenarios: normal conditions, fire detection, intrusion detection, and dual threat conditions. The results show that the system achieved an overall accuracy of 96.5% with detailed performance of 96% for the fire sensor, 94% for the gas sensor, and 98% for the PIR sensor. The average response time was under 300 milliseconds for all types of detection, meeting the real-time system requirements. The decision tree showed excellent classification performance with an F1-score ranging from 95-97% for all categories. The web-based real-time monitoring dashboard successfully displayed sensor status with auto-refresh every 1 second and a data loss rate of only 0.8% during continuous operation.

Streamlit Based Network Intrusion Detection System Prototype with Machine Learning Algorithm

2025-12-16T16:06:48+07:00

Computer network security has become a crucial elemen in the digital era, with the increasing risk of attacks that could potentially disrupt systems and access critical data. An Intrusion Detection System (IDS) powered by Machine Learning is one effective way to automatically detect suspicious network activity. This study aims to create a prototype of a network Intrusion Detection System using Streamlit that applies Machine Learning algorithms, including Naïve Bayes and Random Forest, to classify normal network activity as an attack. The method used in this study is a quantitative approach with an experimental design utilizing a public dataset of labeled network traffic. The research process includes the stages of initial data processing, feature selection, model creation, performance evaluation, and implementation of the Streamlit interface. Test results show that the Naïve Bayes algorithm has the best performance, with an accuracy level reaching 0.8000, an error rate of 0.2000, and an F1 Score of 0.7273. Random Forest recorded an accuracy level of 0.7333, an error rate of 0.2667, and a lower F1 Score of 0.3333. These findings demonstrate that Naïve Bayes is more effective at detecting intrusions and recognizing anomalous network traffic patterns. The Streamlit based system implementation successfully provides an interactive and userfriendly interface, allowing users to perform analysis and understand classification result without in-depth technical expertise. Given the foregoing, the network intrusion detection system prototype built with Streamlit and a Machine Learning algorithm is considered suitable as a simple, informative, interactive, and efficient network security support tool. This research paves the way for future developments, such as the implementation of Deep Learning models and the integration of live network monitoring.

Analysis of the Causes of Multifilament Thread Defects in PT X Using Seven Tools

2025-12-17T15:22:17+07:00

PT X, a leading plastic packaging manufacturer in Indonesia, faces the problem of high defect rates in the production of multifilament yarns that hinder output optimization. This research aims to analyze and improve product quality using the Seven Tools method. Data for the May-October 2025 period shows a total production of 214,045.5 kg with an accumulated defect of 5,029 kg. The results of the Pareto analysis identified two main defects (vital few), namely brittle yarn (38%) and easily broken yarn (31%). Analysis via the control map (p-chart) showed the process was in an uncontrolled condition, especially in August which exceeded the upper control limit (UCL 0.0242). Based on the fishbone diagram, the root cause of the problem comes from the instability of the engine temperature (godet), operator negligence, non-standard SOP, and variations in material quality. To overcome this, it is recommended that companies carry out routine machine maintenance (PPM), install automatic temperature monitoring systems, standardize SOP, and hold periodic training for operators to create process stability and minimize defects on an ongoing basis.

Design of a Software Requirements Specification for a Parental Partnership Assistance Management System in Elementary School in Malang Regency

2025-12-18T13:42:04+07:00

This research is motivated by the need for a more efficient and transparent aid management system in primary education institutions, particularly at SDK Yos Sudarso Kepanjen, where the processes of application, verification, and reporting are still conducted manually. Such conditions lead to various issues, including delayed distribution, data duplication, and difficulties in monitoring assistance. To address these problems, a Software Requirements Specification (SRS) document was designed as a reference for developing a web- and mobile-based Parent Partnership Aid Management Information System. This study employed a system engineering approach consisting of three main phases: analysis, design, and implementation. These phases include problem identification, system workflow redesign, and the development of an initial user interface prototype using HTML and CSS (Bootstrap framework). The results indicate that the SRS document successfully defines the system’s functional and non-functional requirements, including user authentication, aid application, digital verification, automated reporting, and a GPS-based needs mapping feature. It is expected that this SRS document can serve as a guideline for developing collaborative, efficient, and accountable educational information systems in the future.

Cybersecurity Awareness: A Literature Review on Internet Users' Awareness and Safe Behavior

2025-12-18T13:42:41+07:00

The rapid development of information technology has facilitated various aspects of human life, from communication and education to financial transactions. However, this progress has also been accompanied by the growing threat of cybercrime, such as data theft, hacking, and digital fraud. One of the most influential factors contributing to this growing threat is the low level of cybersecurity awareness among internet users. This article aims to review the various literature related to cybersecurity awareness and user safety behavior in the online world. The method used is to review literature from various scientific sources from 2022 to 2025 that discuss cybersecurity awareness, behavior, and education. The results of the study show that while security technology continues to evolve, human awareness remains the weakest point in cyber defense. Therefore, improving education and digital culture is a key strategy in developing safe behavior among internet users.

Classification of Spending Segmentation in Mobile Game Applications Using Random Forest and Decision Tree Algorithms

2025-12-19T22:38:37+07:00

This research aims to classify spending segmentation in mobile game users using Random Forest and Decision Tree algorithms. The dataset consists of demographic attributes, gameplay behavior, session frequency, and historical spending records. Several preprocessing steps uwere applied, including missing value handling, label encoding, one-hot encoding, and feature scaling. The data were divided into an 80:20 training-testing ratio, and hyperparameter tuning was performed using GridSearchCV. The results indicate that Random Forest achieved higher accuracy compared to Decision Tree, demonstrating better generalization for multiclass segmentation (Low, Medium, High spenders). This study shows the potential of machine learning in predicting user spending behavior to support data-driven monetization strategies in mobile game applications.

Administrative and Field Risk Analysis in the PGN Sales Division Using the FMEA Method

2025-12-19T22:39:24+07:00

This study analyzes administrative and field operational risks in the Sales Operation Regional (SOR) III of PT Perusahaan Gas Negara (PGN) using the Failure Mode and Effect Analysis (FMEA) method. The objective of this research is to identify potential failure modes in customer service processes and determine risk priorities that may affect service quality and operational safety. Data were collected through observations, interviews, and documentation during an internship period, covering administrative activities and gas pipeline installation processes. The analysis shows that potential failures are concentrated in three main stages: customer data input, data verification at the head office, and gas pipeline installation. The results indicate that the data input stage has the highest risk level, with the loss of customer data forms recording the highest Risk Priority Number (RPN) value. Technical constraints during data verification and safety-related issues during gas installation were also identified, although with relatively lower RPN values. Overall, the application of the FMEA method provides effective insights for prioritizing corrective actions and improving the reliability, safety, and efficiency of natural gas service operations.

Design and Implementation of a Web-Based Currency Converter System Using an Application Programming Interface

2025-12-19T23:33:22+07:00

This study aims to design and implement a web-based currency converter application that utilizes an Application Programming Interface (API) to provide real-time and accurate exchange rate data. The increasing intensity of global economic activities has created a growing need for fast and reliable currency conversion, while manual conversion methods are prone to errors and data inconsistencies. This research employs the Research and Development (R&D) approach using the waterfall development model, which includes requirement analysis, system design, implementation, testing, and maintenance. The developed application provides two main features: an exchange rate calculator that performs automatic currency conversion based on real-time data, and a currency exchange history feature that presents exchange rate trends in graphical form within a selected period. Testing results indicate that the application runs reliably, delivers fast responses, and consistently displays up-to-date exchange rate information. In conclusion, the proposed application serves as an effective web-based solution for accessing accurate currency exchange information to support international financial activities.

Comparative Performance Analysis of BERT and RoBERTa for Email Spam Classification

2025-12-19T23:34:15+07:00

The rapid advancement of information technology has increased the use of email as a primary digital communication medium, while also contributing to the growing volume of spam emails that threaten productivity and information security through phishing and malware. An accurate and adaptive email spam classification system is therefore required. This study aims to analyze and compare the performance of BERT and RoBERTa transformer models for email spam classification. An experimental research approach was employed using an email dataset consisting of spam and non-spam (ham) classes. The research process includes data collection, text preprocessing, model fine-tuning, and performance evaluation using accuracy, precision, recall, F1-score, and confusion matrix metrics. The results show that both BERT and RoBERTa achieve high classification performance. However, RoBERTa demonstrates superior results, particularly in terms of spam recall and overall accuracy, indicating a stronger ability to detect spam emails. This advantage is attributed to RoBERTa’s optimized pre-training strategy, which improves contextual semantic understanding of email content. In conclusion, RoBERTa is more effective than BERT for email spam classification and can serve as a reliable model for developing robust transformer-based spam detection systems.

Public Opinion Sentiment Analysis of Government Fuel Purchasing Policy by the Private Sector Using Support Vector Machine (SVM) Methods

2025-12-20T19:07:47+07:00

Government policies that provide opportunities for the private sector to participate in the purchasing and distribution of fuel oil (BBM) have triggered various reactions within society. The diversity of opinions expressed on social media reflects public perceptions of the effectiveness and potential impacts of these policies. This study aims to examine public sentiment toward the government policy by applying the Support Vector Machine (SVM) method. Data were collected from various social media platforms containing public responses to the issue of private sector involvement in fuel purchasing. The analysis process consisted of several stages, including data collection, data preprocessing (comprising cleansing, tokenizing, stopword removal, and stemming), feature extraction using the Term Frequency Inverse Document Frequency (TF-IDF) approach, and sentiment classification using the SVM algorithm. The results show that the SVM algorithm performs well in classifying public opinions into two sentiment categories, positive and negative, with a relatively high level of accuracy. The analysis indicates that the majority of public opinions tend to be negative, driven by concerns over potential price disparities, weakened government oversight, and possible socio-economic impacts. The findings of this study are expected to provide constructive input for the government in evaluating and developing energy policies that are more transparent and oriented toward public interest.

Implementation of MTTR and MTBF for Determining the Average Maintenance Interval of Press Machines at PT X

2025-12-21T22:38:51+07:00

The independent internship program at PT X aims to apply Industrial Engineering concepts in a real industrial environment, particularly in the field of machine maintenance. This study focuses on analyzing the reliability of the press machine at Lane Kiln 2, which plays a crucial role in the ceramic production process. The methods used in this analysis are Mean Time Between Failures (MTBF) and Mean Time To Repair (MTTR) to evaluate the frequency of machine failures and the efficiency of repair time. Based on the data processing results presented in Chapter III, the MTBF value obtained is 703 minutes approximately 12 hours, and the availability of the press machine at Lane Kiln 2 is 93.64%. These results indicate that the machine availability level is relatively high and capable of supporting smooth production operations. However, the findings also suggest that although the repair process has been carried out effectively.

Designing an Online To-Do-List System to Support Student Productivity in Managing Lecture Assignments and Islamic Boarding School Activities

2025-12-21T14:19:34+07:00

This study on designing an online To-Do-List system intended to assist mahasiswa santri (Islamic boarding school students attending university) in increasing productivity in managing academic tasks and pesantren actvities. Mahasiswa santri pften struggle to balance academic responbilities with demanding religious routines. To address this issue, the system is developed in the from of a web-based application that can be accessed flexibly. The main features provided include task recording, schedule reminders, and activity progress monitoring. The system design is developed using the prototype method, allowing users to provide direct feedback throughout the design process. The resulting interface and system flow align with user needs and show that the proposed system has the potencial to help users manage their time more efficiently and foster discipline toward their daily schedules. Thus, this system can serve as a digital solution that supports the balance between academic and religious activities for mahasiswa santri.

Keyword : To-Do-List System, Productivity, Mahasiswa Santri, Web-Based Application, Prototype Method

Risk Analysis and Occupational Safety Control Strategies at Each Workstation in the Production Division of PT XYZ

2025-12-21T22:38:07+07:00

In facing increasingly intense business competition, companies are required to have human resources with optimal performance in order to achieve organizational goals. Employee performance is influenced by various factors, including work discipline and the implementation of Occupational Safety and Health (OSH). Work discipline reflects employees’ compliance with company rules and procedures, while the implementation of OSH plays an important role in creating a safe, healthy, and comfortable working environment, thereby increasing productivity and reducing the risk of workplace accidents. PT XYZ, a manufacturing company engaged in fertilizer production, has implemented OSH programs in accordance with applicable Indonesian regulations, including Law Number 1 of 1970 concerning Occupational Safety and Government Regulation Number 50 of 2012 concerning the Implementation of the Occupational Safety and Health Management System. Effective OSH implementation is carried out through hazard identification and risk assessment to minimize the potential for occupational accidents and work-related diseases. Through practical work activities, students are given the opportunity to apply academic knowledge directly in an industrial environment and to understand work processes, safety culture, and professional work ethics. The synergy between higher education institutions and the industrial sector is expected to create a productive, safe, and sustainable working environment, while also preparing graduates to face the challenges of the professional world.

Design of a Local Server-Based Student Data Collection System at Salafiyah Syafi’iyah Sukorejo Islamic Boarding School

2025-12-22T16:51:21+07:00

Management of dormitory room data in Islamic boarding schools is still carried out manually by recording information in notebooks. This approach has the potential to cause data duplication, inaccuracies in record-keeping, loss of students' room placement history, and slow data retrieval processes. This study focuses on designing a dormitory room data management information system based on a local server, without including the implementation stage. The research methods include requirements analysis, system modeling using Unified Modeling Language (UML), and user interface design. The result of this study is a complete and structured system design that can be used as a reference for system development and implementation at a later stage.

Sentiment Analysis of Public Opinion on Rupiah Redenomination Policy Using Support Vector Machine and SMOTE

2025-12-22T16:55:28+07:00

The government’s planned rupiah redenomination has generated a substantial wave of public opinion across social media platforms. This study aims to analyze public sentiment by examining comments on YouTube and classifying them into two categories: positive and negative. The data are collected through web scraping conducted on December 21, 2025, using the keyword “rupiah redenomination.”
Given the pronounced imbalance between negative and positive opinions, this study applies the Synthetic Minority Over-sampling Technique (SMOTE) to balance the class distribution within the training data. The research pipeline consists of text preprocessing, feature extraction using Term Frequency–Inverse Document Frequency (TF-IDF), and classification using a linear-kernel Support Vector Machine (SVM). Experimental results indicate that the SVM model achieves an accuracy of 88.28%. The application of SMOTE is shown to effectively enhance the model’s ability to identify the minority class, with the recall for positive sentiment reaching 0.71. Furthermore, the analysis reveals that public opinion is predominantly negative (83.93%), reflecting widespread concern regarding the potential economic implications of the policy.

Design of a Web-Based Letter Disposition Information System in the Population Control and Family Planning Sector of the Banyuwangi Regency Social Service

2025-12-24T14:04:47+07:00

This research aims to design a web-based letter disposition information system for the Population Control and Family Planning Sector of the Banyuwangi Regency Social Service. Mail management which is still done manually causes various problems, such as delays in the disposition process, risk of data duplication, and difficulties in searching mail archives. To overcome this problem, this research uses field research methods supported by literature study as a theoretical basis. System development was carried out using the Software Development Life Cycle (SDLC) approach using a prototype model which includes communication stages, rapid planning, system modeling, prototype development, and system implementation. The system designed includes features for managing incoming mail, outgoing mail, digital disposition, and electronic archiving of letters. The design results show that this system is able to increase efficiency, accuracy and transparency in the letter administration process, and is expected to become the basis for system development in the next implementation stage.

Design and Build an Internship Information System at PT. Perkebunan Nusantara IV Regional I Medan Web Based

2025-12-24T22:00:30+07:00

The administrative process for the internship program at PT Perkebunan Nusantara IV Regional I Medan is still carried out conventionally, starting from the registration stage to recording attendance and daily journals. This situation causes various problems such as input errors, delays in the verification process, and difficulties in monitoring the attendance and activities of interns. This study aims to design a web-based internship information system as a solution to improve data management efficiency and reduce existing administrative problems. The method used in this study follows the SDLC approach with the Waterfall model, which includes needs analysis, system design, implementation with Laravel and Tailwind CSS, and testing using the Black Box Testing method. The findings of this study indicate that the developed system can support online registration, location-based attendance, daily journal filling, and participant management by the admin more quickly, accurately, and integrated. Testing shows that all main functions operate according to the specified scenario. This system makes a significant contribution to supporting more efficient and up-to-date internship administration and has the potential for further development to improve the quality of internship services at related institutions.

Comparison of Machine Learning Classification Algorithm Performance for Depressive Symptom Recognition in College Students

2025-12-25T11:19:00+07:00

College students are vulnerable to depressive symptoms due to academic, social, and personal pressures, which can impact mental health and academic achievement. Early detection is necessary to prevent this condition from developing into a more serious condition, but conventional methods often lack objectivity. With the development of artificial intelligence, machine learning classification algorithms offer a more accurate approach to recognizing patterns of depressive symptoms. This study compared the performance of several classification algorithms, namely Random Forest, K-Nearest Neighbor, Logistic Regression, Decision Tree, Naive Bayes, and Support Vector Machine, using a dataset of depressive symptoms in college students. Evaluation was carried out based on accuracy, precision, recall, and F1-score. The results showed that Logistic Regression achieved the best performance with an accuracy of 95.62%. This suggests that selecting the right algorithm can improve the effectiveness of early depression detection systems in college students and support data-driven mental health efforts.

Desain User Interface ‘Disaster Spatial Learning Global Partnership’

2025-12-26T21:15:53+07:00

This article presents the design of the User Interface (UI) for the DSL Global Partnership (DSLGP) system, a Web-GIS–based educational platform developed to support disaster literacy and digital learning. The study focuses on translating the Software Requirements Specification into structured UI components, including the Home page, role-based dashboards, and key transactional workflows such as guest registration, assessment distribution, resource validation, and data import operations. The UI is designed using a user-centered approach, emphasizing clarity, accessibility, and role separation to ensure intuitive navigation for research teams, teachers, students, administrators, and public users. This work provides a concise and systematic overview of the UI structure that can guide subsequent implementation and evaluation stages within the DSLGP development cycle.

An Agent-Based Modeling Approach for Crowd Movement in Confined Spaces

2025-12-27T08:39:28+07:00

This study presents an agent-based modeling approach to analyze crowd movement and evacuation performance in confined spaces. The model simulates individual agents navigating toward a single exit while avoiding collisions under varying density conditions. Three evacuation scenarios were evaluated, consisting of 20, 40, and 60 agents within a confined environment measuring 10 × 8 meters. The simulation was executed using a discrete time step of 0.1 seconds, and performance was assessed based on evacuation time and collision frequency. The results indicate that increasing crowd density significantly affects movement efficiency. The 20-agent scenario achieved an average evacuation time of 6.42 seconds with 95.33 collision events. When the number of agents increased to 40, the evacuation time rose to 6.90 seconds with 391.77 collisions. The highest density scenario, consisting of 60 agents, produced an average evacuation time of 7.08 seconds and 890.73 collision events. These findings demonstrate that higher density levels lead to a disproportionate increase in interaction intensity and congestion, resulting in reduced evacuation efficiency. The study confirms that agent-based modeling is an effective approach for analyzing crowd dynamics in confined environments and provides a reproducible framework for evaluating evacuation performance under varying density conditions.

Design of a Server Application for Web-Based Automation of Darul Qur'an Student Registration

2025-12-27T08:38:47+07:00

Study This discuss design server aplication for help automation of the registration process students at Darul Qur’an through a web-based platform. The systemThis developed for increase speed procesissing, accuracy record keeping and neatness previous data management handled manually. Stages study covering requirements analysis, system architecture design, and implementation using web technology. The design result show that the aplication is capable of handling registration in a structured manner, storing data centrally, and offering an easy-to-use interface. This system is expected to improve administrative efficiency and serve as a basis for developing additional features in the next phase.

Design and Development of a Hybrid E-Portfolio Application for Islamic Boarding School Students

2025-12-27T16:16:58+07:00

The development of information and communication technology has encouraged educational institutions, including Islamic boarding schools (pesantren), to adopt digital systems for academic management and documentation. One important requirement in pesantren environments is the management of student portfolios that record academic achievements, religious activities, organizational involvement, and community service. However, limited internet access, digital device usage policies, and uneven technological infrastructure hinder the effective implementation of fully online e-portfolio systems. This study aims to design and develop a Hybrid E-Portfolio Application for pesantren students that can be utilized in both offline and online modes. The system was developed using the System Development Life Cycle (SDLC) with a Prototype model, which includes requirement analysis, initial design, prototype development, and user evaluation. The results indicate that the developed system is capable of supporting structured recording, management, and storage of student portfolios, as well as enabling data synchronization when an internet connection is available. The proposed hybrid e-portfolio application provides an adaptive solution to documentation challenges in pesantren environments and supports sustainable digital transformation in pesantren-based education.

Analysis of user experience with TikTok Shop features among Generation Z in Balikpapan

2025-12-29T13:06:20+07:00

This study was conducted to determine the usability level of TikTok Shop features among Generation Z in Balikpapan. The main focus of this study was how easy these features were to use for the user experience when shopping on the TikTok Shop application. The method used was a quantitative approach using the System Usability Scale (SUS) instrument. Data was collected by distributing questionnaires to 27 respondents who were Generation Z aged 13 to 28 years old residing in Balikpapan. The SUS questionnaire consisted of 10 questions using a Likert scale of 1 to 5, which were then processed to obtain a usability score. The results of data processing showed that the total SUS score obtained was 2,230 with an average value of 82.59. This value indicates that the TikTok Shop feature has a high level of usability. In general, respondents felt that the interface was easy to understand, supported shopping activities, and was in line with Generation Z's social media usage habits. Based on these results, the researcher concluded that the TikTok Shop feature was well-designed and provided a positive experience for Generation Z in Balikpapan.

Workforce Requirement Analysis Using Workload Analysis with a Full-Time Equivalent (FTE) Approach in the Planning & Control Division of PT PLN (Persero) Pusharlis UP2W VI

2025-12-29T13:05:36+07:00

This research was conducted at PT PLN (Persero) Pusharlis Unit Pelaksana Pemeliharaan Wilayah VI (UP2W VI), focusing on the Planning and Control Division. This study aims to analyze the suitability between workload and the number of employees and to determine the ideal workforce requirement to improve operational effectiveness. The method applied is workload analysis using the Full Time Equivalent (FTE) approach through the calculation of task completion time and annual effective working hours. The results indicate differences in workload levels among teams, where some teams experience overload conditions, while others operate under near-optimal workload conditions. The application of the Full Time Equivalent (FTE) method provides a quantitative overview of actual workload conditions and serves as a basis for determining and adjusting optimal workforce formation to support work efficiency in the Planning and Control Division of PT PLN (Persero) Pusharlis UP2W VI

Nutriquest: Diet Therapy Education Through Third-Person Shooter Game

2025-12-30T15:13:38+07:00

Nutrition education plays a crucial role in maintaining health and preventing disease; however, conventional methods such as lectures tend to be passive and unengaging, especially for adolescents. This study aims to develop an interactive educational game as an alternative medium for delivering diet therapy education. This research employed a Research and Development (R&D) approach using the Software Development Life Cycle (SDLC) with a waterfall model, consisting of analysis, design, implementation, and testing stages. The game, NutriQuest, was developed for the PC platform in the third-person shooter genre and is designed based on the Mechanics, Dynamics, and Aesthetics (MDA) Framework to integrate learning content with gameplay. Players explore the game environment, interact with food objects to obtain nutritional information, and complete quizzes to evaluate learning outcomes. Application testing included black box testing, learning media validation, material validation, and user testing. Results showed that all system functions operated properly. Learning media validation achieved a score of 94.7%, while material validation achieved a score of 68.9%, indicating suitability for educational use. User testing involving 23 adolescents yielded a positive response with a score of 91.30%. These findings indicate that NutriQuest is feasible and effective as a game-based learning medium for diet therapy education and can increase user engagement and understanding.

Smartphone Rating Classification Based on Technical Specifications Using Naive Bayes and Feature Importance

2025-12-30T15:14:03+07:00

Classifying smartphone ratings based on technical specifications is crucial for market analysis and consumer electronics marketing strategies. This study applies the Naive Bayes algorithm to categorize smartphone ratings into low, medium, and high levels, while identifying influential features via Mutual Information scores. The dataset includes 1,159 smartphones with eight categorical predictors: screen size, RAM capacity, pixel density (PPI), battery capacity, water resistance, display type, IP rating, and Android OS version. Data preprocessing involved handling missing values with mode imputation, Label Encoding for categoricals, and stratified 80:20 train-test split to preserve class balance. The model achieved 77.16% accuracy and a weighted F1-score of 0.7724 on test data, with 5-fold cross-validation yielding a mean accuracy of 76.53%.Mutual Information analysis ranked RAM capacity (32.51%), PPI (25.78%), and display type (21.09%) as top features. These findings highlight key specs for differentiating rating groups, aiding product analysis, market positioning, and smartphone marketing in Indonesia.

Implementation of a Best Employee Assessment System Using Fuzzy Multiple Attribute Decision Making

2025-12-30T15:13:09+07:00

Employee performance evaluation is a crucial process in human resource management to objectively identify individuals with superior performance. At PT Labari Sehat Perkasa, the evaluation process has traditionally been conducted manually, making it prone to subjectivity, inconsistency, and evaluator bias. These issues often lead to less optimal results in supporting managerial decisions such as promotions and rewards. Therefore, this study implements a technology-based employee evaluation system using the Fuzzy Multiple Attribute Decision Making (FMADM) method. This method was chosen for its ability to transform quantitative data into flexible linguistic values, thus enabling a more objective classification of employee performance. The system applies five main criteria: discipline, responsibility, initiative, job performance, and teamwork, each weighted according to its importance. The implementation results show that 82% of employees are categorized as “Good,” 4% as “Very Good,” and 12% as “Fair.” In conclusion, the application of the FMADM method improves the objectivity of evaluations, simplifies decision-making processes, and enhances employee motivation and productivity at PT Labari Sehat Perkasa.

Cyber Threat Analysis to Personal Data in the Digital Era

2026-01-24T22:59:17+07:00

The development of information technology provides convenience in daily life, but at the same time increases the risk of cyber threats to users' personal data. This study aims to analyze the level of user vulnerability to digital threats and the factors that influence personal data security, including the role of digital literacy and user behavior. The method used is descriptive qualitative, collecting data through a literature review, structured observation of digital device usage patterns, and semi-structured interviews with a number of selected users. The results show that weak password management practices, the tendency to share personal information, and low awareness of suspicious links or applications are the main causes of data vulnerability. The level of attention to vulnerability evaluation, security awareness, and security audits remains low, even though users are aware of potential threats. These findings confirm that digital literacy and user behavior play a crucial role in reducing risks. Therefore, improving cybersecurity education, implementing safe digital practices, and conducting regular security audits and evaluations are necessary to effectively protect personal data and build a safer digital culture.

Music Genre Classification Application Based on Audio Features with Ensemble Learning Algorithm

2025-12-29T13:04:29+07:00

With the exponential growth of digital music, manual genre-labeling has become ineffective. Consequently, automatic music genre classification is crucial for data management and recommendation systems. This research aims to develop an accurate music classification application by comparing individual machine learning models against advanced Ensemble Learning techniques. The methodology involved extracting 26 audio features from the GTZAN dataset, followed by training and hyperparameter tuning ten models, including Random Forest, SVM, XGBoost, and LightGBM. The findings demonstrate that ensemble methods significantly outperform individual models. The highest performance was achieved by a Voting Classifier, which combines the predictive strengths of SVM, XGBoost, and Logistic Regression, reaching a final test accuracy of 72%. This superior ensemble model was then successfully implemented into an interactive web application using Streamlit, proving that this approach is not only highly accurate but also functional for real-time, practical applications.

Information System Security Risk Analysis Using the Annual Loss Expectancy (ALE) Method (Case Study: Website of the Information Systems Department, UPN “Veteran” Jawa Timur)

2025-12-30T15:14:33+07:00

The development of information technology in higher education institutions poses significant security risks to digital assets, including the Department of Information Systems website at UPN “Veteran” Jawa Timur. This study aims to identify, analyze, and evaluate information security risks using the quantitative Annual Loss Expectancy (ALE) method. This method measures risk based on the parameters of Asset Value (AV), Exposure Factor (EF), and Annualized Rate of Occurrence (ARO). The analysis was conducted on four main risk categories: service disruption, device damage, data loss, and system security threats. The results of the study show that information system security threats have the highest potential loss of IDR 81,750,000 per year. After simulating mitigation measures, the annual loss value (ALE Projected) decreased dramatically in all categories. The investment feasibility evaluation using Return on Investment (ROI) resulted in a ratio above 2:1 for all categories, with the highest value of 3.10 in handling security threats. This shows that the proposed security investment is very feasible to implement in order to ensure the continuity of academic services and protect the department's information assets.

Comparison of the Efficiency of Bubble Sort and Insertion Sort Algorithms in Sorting Sales Data in Digital Printing

2025-12-30T15:12:14+07:00

The efficacy of the Bubble Sort and Insertion Sort algorithms is analyzed and compared in this study. arrange sorting algorithms in order of how well they perform at analyzing data from digital printing sales. The rate at which data is processed is crucial for information systems in the age of digital revolution, particularly in sectors where transactions must be completed quickly. The selection of the sorting algorithm has been demonstrated by prior research to significantly affect system performance. According to research by Mifortekh (2023), when it comes to arranging sales data for MSMEs, Insertion Sort outperforms Bubble Sort. Furthermore, research by Iskandar (2020) and Vierdansyah et al. (2023) shows that Insertion Sort uses less memory and doesn't need as much data exchange as Bubble Sort. Global Scientific Journal (2022) and Sabah et al. (2023) both acknowledge the steady performance of Insertion Sort in random data sorting, noting that it continues to perform consistently even with increasing dataset size. Using a quantitative approach with secondary data and Python-based simulations, this study examines execution time in an effort to reinforce empirical evidence supporting the supremacy of Insertion Sort over Bubble Sort and help information system developers choose the best sorting algorithm.

Iris Center Localization Based Hough Transform on Eye Image

2025-12-30T15:12:38+07:00

Localization iris center localization is stage crucial to the system biometrics eyes , tracking view (gaze tracking), as well analysis health based image . Research This propose method detection iris -based center Hough transform on image eye with channel work that emphasizes robustness to noise, variations lighting , and differences iris size . Stages used covering conversion image colored to grayscale for simplify information intensity , noise reduction using Gaussian Blur so that the iris edges are clearer stable , detection candidate iris circle using Hough Circle Transform with setting certain parameters , as well as , determining iris center based on coordinate circle best Hough voting results . Method performance evaluated in a way quantitative through error distance center (center localization error) to ground-truth and in general qualitative through visualization circle results detection on the image . Expected results show that Hough's transformation is capable of give estimate consistent iris center in the image with condition moderate until complex , especially when the iris border is sufficient contrast to sclera . Research This contribute as relative approach simple , fast , and easy implemented For need localization iris center , at the same time provide base development advanced like adaptive iris segmentation and integration with method learning deep For increase resistance to conditions lighting extreme and occlusion by the eyelids or hair eye .

Lean Service Analysis to Identify And Minimize Waste in PT XYZ Non-Production and Investment Procurement Process using Value Stream Mapping

2026-01-15T10:55:33+07:00

The non-production and investment procurement process plays a crucial role in supporting the smooth operation of a company. However, this process often faces problems in the form of waste that causes lengthy processing times. This study aims to identify the dominant types of waste and develop proposals for improving the non-production and investment procurement process at PT XYZ using a lean service approach. The methods used include Value Stream Mapping to map the current process conditions, a questionnaire to determine critical waste, a Fishbone Diagram to analyze the root causes of waste, and the 5S method to formulate improvement proposals. The results show that the dominant critical wastes in the procurement process are waiting, transportation, and inappropriate processing. The Process Cycle Efficiency (PCE) value in the current state is 39.0% and increases to 49.39% in the proposed state (future state), indicating potential for increasing process efficiency. The formulated improvement proposals are expected to minimize waste and increase the effectiveness of the non-production and investment procurement process at PT XYZ.

Application of the Monte Carlo Method in Predicting Customer Frequency in Micro-Scale Laundry Business

2025-12-30T15:09:59+07:00

The main challenge for micro-scale laundry businesses is the variability and uncertainty in daily customer frequency, which directly impacts operational efficiency and resource planning. This study aims to model and predict the average customer frequency using the Monte Carlo Simulation. One hundred days of historical data were processed to determine the customer arrival probability distribution. The simulation was run for 5,000 iterations, resulting in a predicted average daily customer frequency of 20.57 customers/day. This value is proven valid and slightly above the historical expectation. The conclusion indicates that the Monte Carlo Method is an effective tool for modeling systems with high uncertainty. This quantitative prediction provides a strong basis for the business owner to effectively optimize employee scheduling and inventory management of raw materials.

Warehouse Layout Evaluation using the Activity Method Relationship Chart to Maximize Capacity and Efficiency Material Storage

2025-12-31T09:46:54+07:00

Partner companies from various industrial sectors often face issues related to warehouse operational efficiency, particularly in managing suboptimal facility layouts. At PT PLN Nusantara Power UP Paiton, this issue is evident in the inefficient flow of materials within the warehouse, resulting in operational delays and increased costs. This report aims to evaluate the warehouse layout using the Activity Relationship Chart (ARC) method to maximize capacity and material storage efficiency. The ARC method is used to analyze the proximity relationships between facilities in a warehouse based on existing operational processes. The evaluation is conducted by comparing the initial layout with proposed alternative layouts, with a focus on optimizing material flow, reducing moving distances, and increasing warehouse operational efficiency. The analysis results show that the proposed alternative layout can improve warehouse space efficiency, reduce material movement distance, and accelerate workflow. The application of the ARC method successfully identified areas for improvement to increase productivity and reduce material management time. Evaluation of warehouse layout using the ARC method can significantly contribute to improving operational efficiency and better warehouse management at PT PLN Nusantara Power UP Paiton. It is recommended to continue implementing this method in improving the layout of other facilities to support smooth operations and reduce company costs.

Modeling and Simulation of Minimarket Cashier Queues to Reduce Customer Waiting Time at Indomaret Tuntungan I

2025-12-31T09:47:33+07:00

Customer satisfaction in the modern retail industry is significantly influenced by the efficiency of cashier services. Long waiting times can lead to dissatisfaction and potential customer loss. This research aims to analyze and optimize the cashier queuing system at Indomaret Tuntungan I using a discrete event modeling and simulation approach. Primary data on customer arrival intervals and service times were collected through direct observation. A simulation model was constructed using Arena software to replicate the existing system, which operates with two cashiers and separate queues. Several optimization scenarios were proposed and tested, including adding cashiers and implementing an integrated single queue system. The key performance metrics measured were average customer waiting time, queue length, and cashier utilization. The simulation results demonstrate that the combined scenario, featuring a single queue with one additional cashier activated during peak hours, yielded the most significant improvement. This configuration successfully reduced the average waiting time by 67.4% to 2.55 minutes, decreased the maximum queue length to 2.8 people, and achieved a balanced cashier utilization rate of 68.7%. The findings provide a data-driven recommendation for minimarket management to enhance service speed through a synergistic approach of queue restructuring and dynamic resource allocation..

Implementation of ISMO – Based Maintenance System for Shaker Machines at PT X

2026-01-02T10:08:02+07:00

This study examines the implementation of the ISMO maintenance system (Inspection, Small Repair, Medium Repair, and Overhaul) for tobacco screening machines at PT X, located in East Java. The research compares the existing preventive maintenance approach with the ISMO method to evaluate differences in downtime and availability values. The screening machine, developed internally by PT X's engineering team, operates with vertical and horizontal movements to separate non-tobacco materials during processing. Using repair complexity analysis, the study determined the screening machine has a complexity value of 9 as a machine tool. Results indicate that the current company maintenance system achieves 99.86% availability with a total downtime of 580 minutes, whereas the ISMO method results in 97.89% availability with 8,856 minutes of downtime. The ISMO approach requires 3 workers for inspection, 8 for small repairs, 10 for medium repairs, and 11 for overhaul activities. While both methods show excellent availability above 95%, implementing ISMO significantly increases downtime, potentially causing production delays. This study provides a detailed classification of maintenance activities for each ISMO level, offering practical guidance for systematic machine maintenance implementation.

Sentiment Analysis of Public Opinion on the Pertalite Fuel Issue in YouTube Comments Using the Naïve Bayes Method

2026-01-02T10:07:33+07:00

The issue of Pertalite fuel (BBM Pertalite) has generated widespread public reactions on social media, particularly on the YouTube platform, where users actively express opinions through comment sections. This study aims to analyze public sentiment toward the Pertalite fuel issue based on YouTube comments using a text mining approach and the Naïve Bayes classification algorithm. The dataset consists of approximately 3,000 YouTube comments collected via the YouTube Data API and processed through several text preprocessing stages, including cleaning, normalization, tokenization, stopword removal, and stemming. Sentiment labeling was performed using a lexicon-based approach, followed by feature extraction using Term Frequency–Inverse Document Frequency (TF-IDF). The data were divided into training and testing sets with an 80:20 ratio. Experimental results indicate that the Naïve Bayes model achieved an accuracy of 69.67%, with negative sentiment dominating public discourse both in terms of comment frequency and user engagement measured by likes. These findings suggest a prevailing public dissatisfaction with the Pertalite fuel issue and highlight the usefulness of social media–based sentiment analysis as a data-driven instrument for understanding public perception. The results of this study provide valuable insights that can support the evaluation of energy policies and demonstrate the potential of sentiment analysis in policy-related public opinion studies.

Application of Diabetes Risk Prediction Using Machine Learning Algorithms

2026-01-02T23:05:39+07:00

Diabetes mellitus is a chronic disease that poses a significant global health burden, requiring effective early detection strategies to reduce complications and mortality. In recent years, machine learning techniques have been widely applied to support medical decision-making, particularly in disease risk prediction. This study aims to compare the performance of several machine learning algorithms for diabetes risk prediction and to implement the best-performing model into a web-based application. The PIMA Indians Diabetes Dataset was used in this study, and data preprocessing was conducted to address class imbalance and improve model performance. Five classification algorithms were evaluated, namely Logistic Regression, Support Vector Machine (SVM), Random Forest, K-Nearest Neighbors (KNN), and Naive Bayes. Model performance was assessed using accuracy, recall, F1-score, and Area Under the Curve (AUC), with a particular emphasis on recall and F1-score due to their importance in medical screening applications. Experimental results show that the SVM model outperformed the other algorithms, achieving higher recall, F1-score, and AUC values. The selected model was then implemented into a web-based application using the Streamlit framework, enabling users to input clinical parameters and obtain real-time diabetes risk predictions. The results indicate that machine learning models, particularly SVM, can effectively support diabetes risk prediction and demonstrate the potential of integrating predictive models into practical healthcare applications.

Development of an Academic Application Server for Managing Student Data and Grades Based on Web Services

2026-01-02T23:06:13+07:00

The development of information technology encourages higher education institutions to digitalize academic systems in order to improve the efficiency and accuracy of managing student data and academic grades. However, in practice, many institutions still rely on manual methods or separate applications, which leads to data synchronization issues, delays in accessing information, and a high potential for input errors. This study aims to design and develop an academic application server based on web services that is capable of managing student data and grades in an integrated, flexible, and real-time manner.

The system development method used in this research is Rapid Application Development (RAD), which emphasizes fast development through interactive processes, prototyping, and direct user involvement. The research stages include requirement planning, design and development of the API server prototype, implementation of RESTful web services, as well as system testing and evaluation.

The results show that the developed academic application server is able to provide stable services for managing student and grade data, is easily integrated with other applications, and has a faster response time compared to previous manual systems. User Acceptance Testing (UAT) and Black Box testing indicate that all main functions operate as expected without significant errors. In addition, the server architecture is modular, allowing further development, including integration with mobile-based applications. Thus, this study proves that the implementation of web services using the RAD method is effective in supporting efficient, accurate, and integrated academic data management in higher education environments

Optimization of Job Training Through the Implementation of the ADDIE Model in the Safety Academy Program at PT XYZ

2026-01-14T15:29:16+07:00

This study examines the optimization of job training through the application of the ADDIE instructional design model within the Safety Academy program at PT XYZ. The research highlights issues related to the effectiveness of occupational safety training, particularly in material planning, delivery methods, and evaluation of training outcomes. By applying the ADDIE model, comprising Analysis, Design, Development, Implementation, and Evaluation—the training program was structured to enhance employee competencies in health and safety. The findings demonstrate that the ADDIE-based approach contributes to more systematic, relevant, and needs-oriented training, thereby improving both individual performance and organizational outcomes.

Modeling and Simulation of Service Turnaround Time Using a Simple Linear Regression Method on a Discrete Service System

2026-01-03T22:47:19+07:00

Modeling and simulation are important approaches in discrete service system analysis to understand the behavior of systems as well as estimate their operational performance. One of the main problems in service systems is the uncertainty of service completion time which is affected by various operational parameters. This study aims to build a computational model in predicting service completion time using a simple linear regression method as a mathematical approach based on historical data. The independent variable used represents the system load, while the dependent variable is the duration of service completion. The linear regression model is constructed through a mathematical modeling process and simulated using actual data to generate an estimated service time. The simulation results were then analyzed using prediction error metrics, such as Mean Absolute Error (MAE) and Mean Absolute Percentage Error (MAPE), to evaluate the model's performance. The results show that simple linear regression is able to provide a consistent and fairly representative service time estimate in discrete service systems with low complexity. This approach can be used as a basis for operational decision-making as well as a starting model for the development of more complex computing-based prediction systems.

Image Enhancement of Palmprint Images Using High-Pass Filter and Fast Fourier Transform Methods

2026-01-04T18:59:44+07:00

This study investigates the effectiveness of High-Pass Filter (HPF) and Fast Fourier Transform (FFT) techniques for enhancing palmprint image quality. The methodology encompasses preprocessing stages including image cropping, resizing to 256×256 pixels, grayscale conversion, and histogram equalization. Enhancement is subsequently performed using spatial-domain HPF with two coefficient variations (K=1 and K=0) and frequency-domain FFT with three distinct high-pass filters: Ideal High-Pass Filter (IHPF), Butterworth High-Pass Filter (BHPF), and Gaussian High-Pass Filter (GHPF). Experimental evaluation of 30 palmprint image samples utilizes Mean Square Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) metrics. Results demonstrate that HPF with K=1 achieves superior performance with average MSE of 7.064544 dB and PSNR of 40.01314 dB. Among frequency-domain approaches, IHPF yields optimal results with average MSE of 9.354056 dB and PSNR of 38.537046 dB. The research contributes to biometric image processing through comparative analysis of spatial and frequency enhancement methods, with practical implementation via a MATLAB-based graphical interface.

Unsupervised Underwater Images Segmentation Based on Mean Shift Algorithm

2026-01-05T13:01:54+07:00

Digital image processing is an important field in pattern recognition and computer vision, where the process of separating an object with the background acts as one of the crucial roles in executing image processing. This research will talk about mean shift algorithm implementation in processing a set of undersea images to see its effectiveness in separating the background with the object automatically based on pixel distribution and color intensity. How the mean shift algorithm works is by doing a reading process to find the center of a pixel cluster in an image before this algorithm starts to cluster or group pixels with similar characteristics. In this research, a set of image input will be used as a test under mean shift algorithm to give output of an optimal pixel segmentation or grouping, leading to a proof that the mean shift algorithm has the capability to separate a main object from the background, especially in images with intense and high contrast. Even so, in images with less intense and lower contrast, the segmentation is not as accurate. The image input will have to undergo some more pre-processing before the mean shift segmentation is implemented. This research outcome is that the mean shift algorithm is effective for segmenting marine animal images based on colors without having to initialize how many clusters as the conclusion where this method is applicable in many computer vision programs such as object detection or image pattern recognition.

Analysis of the Application of Reliability, Availability, and Maintainability (RAM) Method in CNC Plasma Cutting Machine Maintenance at PT XYZ

2026-01-05T21:40:16+07:00

This study aims to analyze the application of the Reliability, Availability, and Maintainability (RAM) method in the CNC Plasma Cutting machine maintenance system at PT XYZ as one of the main equipment in the production process in the maritime industry, especially in shipyard companies. The analysis was conducted based on machine operational data during a one-year observation period which includes operating time, planned downtime, unplanned downtime, and the number of machine failure events. A quantitative approach was used to calculate the Mean Time Between Failure (MTBF), Mean Time To Repair (MTTR), and Availability parameters as indicators of machine maintenance performance. The analysis results show that the CNC Plasma Cutting machine has an MTBF value of 103.2 hours, an MTTR of 2.54 hours, and an availability of 98.71%. These values indicate that the machine has a high level of reliability and availability, and a maintenance system that is able to restore the machine's condition in a relatively short time after a disruption occurs. However, unplanned downtime is still found which has the potential to affect the effectiveness of machine operations. Overall, the application of the RAM method is able to provide a comprehensive picture of machine maintenance performance and can be used as a basis for evaluation in preparing more targeted and data-based maintenance policies to support the continuity of the production process in shipbuilding companies.

Loan Default Risk Prediction System in Online Loan Services using Machine Learning and Streamline

2026-01-06T14:45:22+07:00

The rapid development of information technology has driven innovation in the financial sector, particularly in the field of credit lending services. However, the increasing number of credit lending services also leads to a higher risk of default, which can lead to financial losses for lenders. This study aims to develop a loan default risk prediction system using machine learning algorithms, namely Naïve Bayes and K-Nearest Neighbor (KNN), implemented through the Streamlit framework. This study applies a quantitative method with a data mining approach based on the CRISP-DM framework, utilizing the German Credit dataset consisting of variables such as age, occupation, housing, savings account, loan amount, and purpose. The models were evaluated using a confusion matrix to measure accuracy. The results show that the Naïve Bayes algorithm achieved the highest accuracy (86.4%) in predicting loan decisions, followed by KNN (60%). The developed Streamlit-based application provides interactive visualizations, training models, and prediction features, enabling users to assess credit risk efficiently. This system is expected to help financial institutions identify potential defaulters more accurately and improve the overall performance of credit lending services.

Web Based Public Complaint Service System

2026-01-11T22:34:17+07:00

This research is motivated by the suboptimal quality of public services and the management of public complaints, which are still conventional (manual recording, verbal, or text messages) in many villages in Indonesia, including the Purwodadi Village Hall. Based on observations, the Purwodadi Village Hall does not yet have an integrated system to accommodate, manage, and monitor complaints in a structured manner. This results in complaint data often not being optimally documented and difficult to use as evaluation material. Starting from these problems, this study aims to design and implement a Web-Based Public Complaint Service System at the Purwodadi Village Hall. The development of this system uses the Waterfall Method, which is part of the Software Development Life Cycle (SDLC), including the stages of needs analysis, system design, coding, testing, and maintenance.

Designing a Web-Based E-Counseling Application using the Biopsychosocial Balance Theory to Optimize Academic Guidance and Counseling Services (Case Study: Darussalam Gontor University)

2026-01-07T12:36:30+07:00

Academic guidance and counseling services play an important role in supporting student welfare and academic success. However, the implementation of counseling services at Darussalam Gontor University still faces various obstacles, such as an unstructured scheduling process, manual counseling history recording, and limited access to services for students. This study aims to design and develop a web-based e-counseling application using biopsychosocial balance theory to optimize academic guidance and counseling services at Darussalam Gontor University.The research method used is the System Development Life Cycle (SDLC) with a Waterfall approach, which includes the stages of needs analysis, system design, implementation, testing, and maintenance. The developed application provides features for making counseling appointments, selecting counselors, filling out counseling forms based on biological, psychological, social, and spiritual aspects, and confirmation of counseling schedules by counselors. System testing was conducted using Black Box Testing to ensure that all system functions ran according to user needs. The results of the study show that the e-counseling application that was developed works well and meets the needs of users, including students, counselors, and administrators. With this application, the counseling service process has become more structured, efficient, and accessible. It is hoped that this e-counseling application can become a digital solution that supports the improvement of the quality of academic guidance and counseling services at Darussalam Gontor University.

Artificial Intelligence in Recruitment, A Systematic Literature Review on Trends, Challenges, and Future Directions

2026-01-09T15:34:29+07:00

This study aims to examine the scientific literature on the use of artificial intelligence (AI) in the employee recruitment process, with a focus on current trends, existing challenges, and future development directions. Using a systematic literature review approach, the study filters scholarly articles from reputable databases such as Scopus and Web of Science, applying inclusion criteria that consist of English-language publications, published within the last 20 years (2005–2025), and specifically focused on the use of AI in recruitment. Articles that are duplicated, irrelevant to the topic of recruitment, or not peer-reviewed were excluded from the analysis. The main findings indicate that AI offers efficiency in applicant screening, reduces human bias, and enhances the candidate experience. However, significant challenges also emerge, including algorithmic bias, ethical concerns, and organizational resistance to adopting new technologies. Recent trends point to a shift toward the use of machine learning, recruitment chatbots, and predictive analytics in HR decision-making. This study provides a theoretical contribution by synthesizing and categorizing prior research findings, and a practical contribution for HR practitioners in understanding the potential and risks of AI implementation. It also fills a gap in the literature by addressing the lack of a comprehensive synthesis that systematically maps the development of AI research in recruitment.

Implementation of Microservice Architecture in a Machine Learning-Based Expert System for Sleep Disorder Diagnosis

2026-01-09T12:46:30+07:00

Sleep disorders constitute a significant health concern that frequently remains undiagnosed due to restricted access to adequate clinical assessment facilities. This study aims to develop an accurate and accessible early detection system for sleep disorders by implementing the Gaussian Naive Bayes algorithm within a hybrid microservice architecture. The development methodology involves decoupling the intelligent computing service, built on Python (Flask), from the user interface, developed using PHP (CodeIgniter 3), with communication facilitated via an Application Programming Interface (API). The model was trained utilizing a sleep disorder diagnostic dataset comprising 1,000 medical records and evaluated using the 10-Fold Cross-Validation method. Experimental results indicate that the developed model demonstrates superior classification performance, achieving an accuracy of 97.40% and a recall of 99.66%. The high recall value evidences the system's superior sensitivity in detecting positive cases, thereby effectively minimizing the risk of undetected patients (False Negatives). System integration via API proved stable in delivering real-time diagnostic visualization, confirming that this hybrid architecture offers a valid, modular, and responsive solution for the implementation of intelligent healthcare systems.

Forecasting to Support Company XYZ's Financial Planning in 2025 using the Moving Average, Single and Double Exponential Smoothing Methods

2026-01-09T12:47:02+07:00

XYZ Company derives part of its revenue from training services. The primary reason for creating this study is the challenge of achieving its consistently increasing annual revenue target, necessitating an accurate forecasting strategy to support operational decision making and resource allocation in the program, evaluation, and collaboration departments. The chosen methodology involves a quantitative time-series analysis, including 2-month and 3-month Moving Averages (MA), as well as Single and Double Exponential Smoothing. The historical data analyzed covers revenue from January to November 2025. The results indicate that the 3-month Moving Average (MA) method has the best accuracy with a Mean Absolute Percentage Error (MAPE) of 32% (acceptable accuracy), resulting in a revenue forecast for December 2025 of Rp5,521,000,000. In comparison, the 2-month MA method produced a constant forecast value of Rp5,657,500,000 for the period December 2025 to February 2026, but with a lower level of accuracy (MAPE 82%). The accurate implementation of this forecasting method provides strategic recommendations for XYZ Company in developing more precise financial planning. Although this study does not employ Automated learning models and classic time-series analysis remain widely used for practical financial planning.

Designing the User Interface (UI) for a Mobile-Based Ordering System at Burger Stalls

2026-01-10T19:21:28+07:00

The development of mobile technology has encouraged culinary businesses to improve service quality through the use of digital systems, including in the ordering process. Warung Burger Zidan, as a small-scale culinary business, still uses a manual ordering system that has the potential to cause recording errors, service delays, and order discrepancies. This study aims to design a simple, easy-to-use mobile-based ordering system user interface (UI) that suits the operational characteristics of a burger restaurant. The research method used is qualitative with data collection techniques in the form of literature studies, interviews, and observations. The UI design process was carried out using a User-Centered Design (UCD) approach that placed user needs and characteristics as the main focus of the design. The results of the study are a mobile ordering application interface design that includes a menu display, shopping cart, automatic price calculation, and cash and QRIS payment method options. The design evaluation shows that the resulting UI design is capable of supporting a more structured ordering process, increasing service efficiency, and minimizing the risk of recording errors. This study is expected to be the basis for the development of a digital ordering system for small-scale culinary businesses.

Development of an Educational Game for Introducing Object Names in English using the Multiplicative Random Number Generator Algorithm

2026-01-11T11:20:13+07:00

The acceleration of technology has produced a substantial impact on the education sector, leading to the emergence of various teaching tools based on digital multimedia devices. However, the process of teaching English to first-grade students at SDN Wainggai, East Sumba, particularly for object recognition material, still heavily relies on traditional methods, namely oral instruction and the use of textbooks. This approach has caused students difficulty in memorizing vocabulary and requires a longer time for them to grasp the lesson, proven by their average score of only 67, which is below the school's Minimum Completeness Criteria (KKM 70). To overcome these learning barriers, this research proposes and develops an educational game that operates on the Android platform. Its primary goal is to support educators while simultaneously boosting students' motivation and vocabulary retention. The application presents objects with names in both English and Indonesian, and is designed as an interactive tool for first-grade students. The Multiplicative Random Number Generator (MRNG) algorithm is implemented to support the dynamic random features within the game. The expectation is that this educational game will foster a more enjoyable, engaging, and easily understandable English learning environment, ultimately optimizing students' learning outcomes and memorization skills.

Predicting House Prices in South Jakarta using Linear Regression and Decision Tree Regressor Methods

2026-01-11T11:19:10+07:00

The development of digital technology and data analytics has significantly improved predictive capabilities across various sectors, including the property market. House price prediction is a critical element in decision-making for buyers, sellers, investors, and developers, as prices are influenced by factors such as location, building size, land area, number of rooms, and available facilities. This study aims to build a house price prediction model in South Jakarta using Linear Regression and Decision Tree Regressor machine learning algorithms and compare their performance based on regression evaluation metrics. The dataset consists of 1010 entries, divided into 80% training data and 20% testing data. The experimental results show that Linear Regression produced the best performance with an R² score of 0.7713, meaning that the model can explain 77% of the variance in house prices. The model achieved a prediction error of approximately MAE Rp 1.98 billion and RMSE Rp 3.26 billion. Meanwhile, the Decision Tree Regressor obtained an R² score of 0.5560 with higher prediction errors, indicating a tendency toward overfitting and weaker generalization on testing data. Therefore, Linear Regression is recommended as the most effective approach for predicting property prices in South Jakarta and has the potential to be applied in decision-support systems for real estate market analysis.

Implementation of Serverless Architecture and Real-Time Database on Laprak Tracker Application using ReactJS and Firebase

2026-01-11T11:18:21+07:00

Management of laboratory practicum reports often faces challenges such as inefficient progress tracking and manual data synchronization between students and laboratory assistants. This research aims to develop a web-based application named "Laprak Tracker" to streamline this process. The system is built using a serverless architecture with Google Firebase as the backend and Cloud Firestore for real-time data synchronization. The frontend is developed using ReactJS with a Single Page Application (SPA) approach to ensure responsiveness and high performance. The methodology used is the Waterfall model, encompassing requirements analysis, system design, implementation, and testing. Testing results using the Black Box method show that all features, including authentication, module management, and real-time status updates (ACC tracking), function correctly. This application provides an efficient solution for students to monitor their practicum progress and ensures data integrity through a centralized cloud system.

Design and Implementation of an Android-Based Financial Management Application for the Information Technology Study Program

2026-01-11T15:52:34+07:00

The rapid development of mobile technology has encouraged the use of Android-based applications in various domains, including financial management. Students of the Information Technology Study Program often face difficulties in recording and managing personal income and expenses systematically, highlighting the need for a practical and accessible solution. This study aims to design and implement an Android-based financial management application to support efficient and structured financial administration. The research employs a Design and Creation methodology with a qualitative approach, following the Waterfall system development model that includes analysis, system design, implementation, and testing. The application was developed using Android Studio with the Kotlin programming language, featuring modules for recording income, expenses, assets, and generating financial reports. Functional testing using Black Box Testing indicates that the system operates according to specifications, while usability assessment shows that users can manage finances more effectively and intuitively. The findings suggest that the developed application reduces manual errors, improves data accessibility, and provides a reliable tool for managing personal finances. The application can be further enhanced with advanced features such as data analytics or intelligent recommendations to support decision-making in academic environments.

Development of Interactive Educational Games with Monte Carlo Algorithm to Increase Third Grade Students' Learning Motivation on Plant Ecosystem Material

2026-01-11T16:42:10+07:00

The objective of this research is to develop an Android-based interactive educational game as an innovative solution to enhance learning outcomes among 3rd-grade students at SDN Wainggai in the subject of plant ecosystems. The research background is based on observations indicating that science learning still employs conventional methods, which results in low student interest and learning outcomes. The interactive educational game was chosen for its potential to present material attractively through a combination of text, images, audio, and interaction. The game development utilized the Waterfall method. The plant ecosystem material is presented interactively through conceptual explanations, educational mini-games, and evaluation quizzes. The Monte Carlo algorithm was implemented to randomize the evaluation questions to increase variation and reduce boredom. The research results show that this educational game is proven effective in increasing comprehension and learning outcomes. The results of pre-test and post-test evaluation conducted on 19 students showed a significant improvement. The increase is evident from the rise in the average scores of the initial tests (64.78, 59.0, 55.26) and the final tests (82.68, 86.42, 84.89). The average score increase reached a percentage of 27.63%, 86.39%, and 53.61%. In conclusion, the developed educational game was successfully used as a learning medium to help students understand the plant ecosystem material.

Design of a Web-Based Congregation Complaint Information System (Case Study of GKS Andumanang)

2026-01-12T21:50:52+07:00

Perkembangan teknologi informasi yang pesat telah memberikan dampak signifikan terhadap efektivitas dan efisiensi layanan di berbagai bidang, termasuk pelayanan gereja. Gereja Kristen Sumba (GKS) Andumanang adalah gereja lokal yang aktif dalam pelayanan spiritual dan sosial kepada jemaatnya. Namun, sistem pengaduan jemaat yang masih dilakukan secara manual, seperti melalui pertemuan tatap muka atau telepon, menimbulkan berbagai kendala dalam pengajuan, pencatatan, dan penanganan pengaduan. Jemaat yang tinggal jauh dari gereja dan jadwal yang padat dari masing-masing pihak merupakan kendala utama dalam proses komunikasi dua arah antara jemaat dan gereja. Untuk mengatasi masalah tersebut, penelitian ini bertujuan untuk merancang dan mengembangkan sistem informasi pengaduan jemaat berbasis web yang memungkinkan proses pengaduan dilakukan secara efektif, efisien, dan terdokumentasi dengan baik. Sistem ini dibangun menggunakan metode Waterfall, yang meliputi tahapan analisis kebutuhan, desain sistem, implementasi, dan pengujian. Dengan implementasi sistem ini, diharapkan dapat meningkatkan kualitas pelayanan gereja, memperkuat transparansi dan akuntabilitas, serta mempercepat respons terhadap pengaduan jemaat.

Sentiment Analysis of Mamikos Application User Reviews using the Naive Bayes Classifier Algorithm

2026-01-15T10:51:31+07:00

Mamikos, a boarding house search application, has grown to be among Indonesia's most well-liked sites. User reviews on the Google Play Store provide valuable data regarding user satisfaction. However, the large volume of reviews makes it difficult for developers to analyze sentiments manually. This study seeks to do sentiment analysis on evaluations of the Mamikos application to categorize user thoughts as positive or negative sentiments. The method used is the Naive Bayes Classifier with TF-IDF (Term Frequency-Inverse Document Frequency) feature weighting. The research stages include data collection (scraping), text preprocessing (including cleaning and stopword removal using the Sastrawi library), and classification. The test results show that the Naive Bayes algorithm can classify sentiments with an accuracy of 87.96 %. This research is expected to provide insights for application developers regarding aspects that need improvement based on user complaints.

Comparison of Road Damage Classification Accuracy Based on Grayscale Bit Depth Using GLCM Feature Extraction and Backpropagation Neural Network

2026-01-14T14:10:27+07:00

Road surface damage such as cracks, potholes, and patches can reduce driving comfort and threaten road user safety. Manual road inspection is time-consuming and inefficient, especially for large urban areas. This study proposes an image-based approach to classify road damage types using Gray Level Co-occurrence Matrix (GLCM) feature extraction and a backpropagation artificial neural network. Road images were captured directly in Palembang City using a smartphone camera and converted into grayscale images with five different bit depths: 4-bit, 5-bit, 6-bit, 7-bit, and 8-bit. Texture features including contrast, correlation, energy, and homogeneity were extracted using GLCM and used as inputs to the neural network classifier. Experimental results show that higher grayscale bit depth produces better classification accuracy, with 8-bit grayscale achieving the highest performance compared to lower bit depths. The results confirm that grayscale resolution significantly affects texture representation and classification accuracy. This approach can support automated and efficient road damage detection systems.

Digital Image Data Security Using MATLAB -Based AES-128 Algorithm

2026-01-14T14:10:57+07:00

Digital images are widely used for storing and transmitting sensitive information , so that aspect data security is becoming very important. This research discuss digital image data security using the Advanced Encryption Standard (AES) algorithm with a long key 128 bit implemented in MATLAB environment. The method used is cryptography . symmetric , where the encryption and decryption processes use a key the same secret . The encryption process done to randomize the pixel value of the original image so that visual information cannot be recognized , while the decryption process aim return image encrypted to condition back to the beginning without data changes . System evaluation done through histogram analysis , process time measurement , and evaluation quality image using the Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR) parameters. The test results show that The encrypted image has a random histogram pattern and does not resemble original image . In addition that , the decryption process produce MSE value of zero and PSNR value not finite , which indicates the absence of lost information . Thus , it can concluded that the MATLAB -based AES-128 algorithm is effective and reliable in improving the security of digital image data .

Design and Construction of a Web-Based Work Permit and Leave Application Information System for CV. Dodi Brothers

2025-08-10T21:19:46+07:00

CV Company. Dodi Brothers is a private company engaged in the procurement of goods and services. At this time the company CV. Dodi Brothers does not yet have a website for applying for work permits and leave, so the aim of this research is to make it easier for employees to apply for work permits and leave, and to make it easier for admins to approve work permit and leave applications submitted by employees. The research method uses a qualitative descriptive method with data collection techniques in the form of observation, interviews and literature study, data sources consist of primary and secondary. Meanwhile, the method for system development used is the RAD (Rappid Application Development) method. The system design tool used is UML (Unified Modeling Language). This website is designed using the PHP (Hypertext Processor) programming language. MySQL database and coding using Visual Studio Code.

Development of an Artificial Intelligence-Based Adaptive Typing Training System to Improve Accuracy and Speed

2026-01-15T10:45:25+07:00

Mastering the skill of fast and accurate ten-finger typing is a crucial competency in the era of digital transformation. However, conventional, static typing training methods are often ineffective because they fail to adapt the training material to each user's specific weaknesses. This research aims to develop a Reinforcement Learning-based adaptive typing training system capable of dynamically personalizing training material to improve user typing speed and accuracy. The research method used was an experiment implementing the Q-Learning algorithm, in which an intelligent agent determines training material based on the user's error profile to maximize performance improvement. Evaluation was conducted on 50 students divided into experimental and control groups. System performance was analyzed through the convergence of the agent's learning curve and a comparison of pre-test and post-test results. The results showed that the Reinforcement Learning agent successfully learned the optimal training strategy and achieved reward stability in the final stage of training. User testing demonstrated that the adaptive system was able to increase Words Per Minute by 30–68% and significantly improve accuracy compared to static methods. Thus, the Reinforcement Learning approach has proven effective in creating a typing training system that is adaptive, efficient, and tailored to individual needs.

Development of Multimedia-Based Interactive Learning Media to Improve the Understanding of Fifth Grade Students at Inpres Yubuwai Elementary School

2026-01-15T15:57:39+07:00

This study focuses on developing interactive multimedia learning media to improve the understanding of fifth-grade students at Inpres Yubuwai Elementary School in the subject IPAS, specifically the material on food chains. The media was developed using Unity 3D as an interactive application development platform, combined with the Model Development Life Cycle (MDLC) method to ensure a systematic and structured development process. To increase variety and challenge in learning, the Fisher-Yates Shuffle algorithm was applied to the quiz feature to randomize the order of questions. The application of this algorithm aims to avoid repetitive question patterns and encourage students to think more critically. The developed interactive learning media was then tested directly in a classroom environment. The test results showed a significant increase in student understanding, as indicated by a 71.6% increase in scores from the pre-test to the post-test. This improvement proves that the use of technology-based learning media that integrates interactive elements, games, and question randomization algorithms can effectively support the student learning process. Overall, this study shows that the combination of Unity 3D, the MDLC method, and the Fisher-Yates Shuffle algorithm has great potential in producing learning media that has a positive impact on elementary school students.

Application of K-Means Clustering in Public Opinion Analysis Based on Text Mining on Social Media

2026-01-19T13:23:34+07:00

The development of social media platforms has made Twitter a crucial tool for understanding public opinions on various social and policy aspects. Analyzing patterns of public opinion on large and unstructured text datasets requires the use of efficient computational techniques. This study aims to explore the public views of Indonesian citizens on Twitter by applying text processing methods and k-means clustering. The data used in this research method consists of a collection of Indonesian-language tweets taken from a common dataset. The research process includes data collection, text preparation (including lowercase conversion, word separation, removal of common words, and stemming), and feature extraction through TF-IDF. Then, the k-means clustering algorithm is applied to group tweets based on the similarity of word patterns used. The results of this study show that this approach can create representative groups of opinions and help identify the main themes discussed on Twitter. These findings are expected to contribute to the study of the use of data mining and clustering techniques in social media-based public opinion analysis that has been used in the life of Indonesian society.

Application of the Autoregressive Integrated Moving Average (ARIMA) Method in Forecasting the Number of Young Construction to Support Human Resource Planning in the Construction Sector

2026-01-19T13:28:09+07:00

This study aims to apply the ARIMA method in predicting the number of young construction workers in Indonesia during the period 2012 to 2022. The data source used is from the Badan Pusat Statistik(BPS), which includes data on skilled workers in 34 provinces in Indonesia. In this study, three different ARIMA models were tested, namely ARIMA(1,1,1), ARIMA(0,1,1), and ARIMA(1,1,0), to see which model provided the best prediction results. Based on the evaluation results using Mean Squared Error (MSE) and Mean Absolute Percentage Error (MAPE), the ARIMA(0,1,1) model showed the best performance with the smallest MSE and the lowest MAPE, which was 2.5%. These results indicate that simpler ARIMA models, such as ARIMA(0,1,1), are more effective in predicting the demand for young construction workers, thereby assisting in more targeted training planning, resource distribution, and development policies. This study is expected to make a significant contribution to the development of the construction sector in Indonesia by providing a solid basis for data-driven decision-making in workforce planning in this sector.

Designing a Mobile Application Prototype for Accessing and Publishing Student Journals using Figma

2026-01-20T21:30:33+07:00

The advancement of digital technology has driven significant changes in academic information management, including student journal publications. This study aims to design a mobile application prototype as a medium for accessing and publishing journals within the Faculty of Science and Technology at the State Islamic University of North Sumatra. The method employed is the prototyping model, which emphasizes an iterative process between designers and users. The research stages include needs identification, initial design creation, evaluation, and design refinement using Figma as an interface design tool. The results of this research demonstrate a prototype application that includes key features such as journal uploading, searching, and downloading of student scientific papers. Initial evaluations reveal that the design is easy to understand, has clear navigation, and meets user needs. Furthermore, application testing shows a significant improvement in the effectiveness of journal publication access for students, as well as ease of managing academic papers in a more structured and systematic way. This prototype is expected to serve as a foundation for the development of a more integrated and efficient student journal repository system in the future. Additionally, this study contributes to the advancement of information technology in the academic context, particularly in faculty and higher education institutions in Indonesia.

Predicting the Number of Thai Tourists Visiting Indonesia Per Month Using Multi-Layer Perceptron Backpropagation and Adam

2026-01-28T13:08:34+07:00

This research seeks to forecast the monthly number of Thai tourist arrivals in Indonesia using an Artificial Neural Network (ANN) model based on Multilayer Perceptron (MLP) and the backpropagation algorithm. The historical tourist data is treated as a time series, applying the sliding window technique to use data from the past 12 months as input features and the following month as the target output. To enhance the model's training stability, the data is normalized using Min–Max Scaling. The constructed MLP model includes two hidden layers, with the input layer consisting of 12 neurons, each hidden layer containing 21 neurons, and a single neuron in the output layer. The first hidden layer utilizes the ReLU (Rectified Linear Unit) activation function, while the second layer employs tanh (hyperbolic tangent) to capture more intricate nonlinear patterns. The model is trained using the Adam optimizer with a learning rate of 0.01 over 2000 iterations, aiming to minimize the Mean Squared Error (MSE). The training results indicate that the MLP model effectively learns the data patterns, providing accurate predictions with strong alignment between the predicted and actual values. Predictions for the next 24 months reveal a fluctuating trend in tourist visits, mirroring the dynamic characteristics of time series data. Therefore, this approach proves to be an effective tool for tourism policy formulation and decision-making within the Indonesian tourism industry. The Adam optimizer and tanh activation function contribute to the model's enhanced learning efficiency and stability, especially when handling data with significant fluctuations.

Keywords: Forecasting, Tourists, Thailand, Indonesia, Arrivals, Tourism, Data, Artificial Neural Network, Multilayer Perceptron, Backpropagation, Adam, ReLU, tanh.

Prediction of Domestic Tourist Trips in North Sumatra Using Multilayer Perceptron Neural Networks (MLP)

2026-01-23T14:13:58+07:00

The tourism sector significantly contributes to the economic development of North Sumatra Province. To assist effective planning and policy decisions, reliable predictions of domestic tourist trips are needed. This research applies an Artificial Neural Network (ANN) using a Multilayer Perceptron (MLP) structure to estimate the number of domestic tourist visits to North Sumatra during the period 2019–2024. Considering that tourism data tend to be nonlinear and highly dynamic, the MLP approach was selected because of its ability to learn complex time-based patterns. A sliding window approach was applied, using 12 months as input variables and one month as the target for each forecast. Data preprocessing was conducted using Min–Max normalization to enhance learning performance. The network was trained with the Adam optimization algorithm for 2000 iterations. Several MLP model configurations with different numbers of hidden layers and neurons were compared to identify the most suitable structure. Model evaluation was carried out using Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE), and forecasting was tested over a 24-month period. Among the tested architectures, MLP (12–10–6) achieved the best performance with RMSE = 1,546,774.05, followed by MLP (12–6–2) with RMSE = 1,587,366.76, and MLP (12–32–16) with RMSE = 1,699,310.07. This study demonstrates that ANN, specifically MLP, is a robust tool for predicting tourism demand, offering actionable insights for stakeholders in North Sumatra’s tourism sector to guide sustainable development and resource allocation.

Analysis of a Simple Bayesian Network and its Extensions to Robot Decision Making

2026-01-14T08:32:38+07:00

This paper investigates how probabilistic graphical models and their quantum extensions can support decision making under uncertainty in robot–human interaction. Using a simple person-following task as a representative example, the study compares a classical Bayesian Network (BN) with two quantum extensions: a quantum-enhanced Bayesian Network (QeBN) and a full Quantum Bayesian Network (QBN). The purpose of this paper is to examine the expressive limitations of classical Bayesian decision models in contexts involving ambiguous sensing and partially conflicting evidence, and to evaluate whether quantum-inspired representations can provide richer and more flexible decision mechanisms while remaining interpretable. In the classical BN, uncertainty in human motion and robot actions is represented through conditional probability distributions, and action selection is performed by marginalizing over hidden variables. This framework supports decision making by combining prior beliefs and sensor-based likelihoods in a principled and computationally efficient manner. However, because inference relies on additive probabilities and conditional independence, alternative explanations contribute only as weighted mixtures, preventing interaction between competing hypotheses. To address this limitation, the paper reviews a QeBN in which classical probabilities are lifted to complex probability amplitudes while the original graph structure is preserved. This extension retains classical marginals but allows phase-dependent interference during marginalization, enabling non-additive belief updates that capture contextual effects such as sensor fusion ambiguity or conflicting cues in human–robot interaction. Building on this, a full QBN formulation is reviewed in which beliefs and conditional relationships are represented directly by quantum states and density operators. Inference is performed through joint state construction, quantum marginalization via partial trace, and measurement-based conditioning, providing a fully quantum-native alternative to classical Bayesian reasoning. Through analytical walk-through examples, the paper clarifies how each framework supports decision making, highlights their conceptual and mathematical differences, and demonstrates how quantum extensions can enhance expressiveness and context sensitivity beyond classical Bayesian models. The results position BN, QeBN, and QBN as complementary tools along a spectrum of decision-making models, offering increasing representational power for robotic systems operating in uncertain and human-centered environments.

Air Quality Monitoring System for Hazardous Gas Detection and Fire Risk in a Chemical Laboratory

2026-01-18T18:59:24+07:00

Chemical laboratories require a safe and controlled environment due to the frequent use of hazardous and flammable substances that may pose serious health and fire risks. Inadequate air quality monitoring can increase the likelihood of gas exposure and uncontrolled fire incidents. This study focuses on the development of an air quality monitoring system for detecting hazardous gases and fire risk in a chemical laboratory. The proposed system is based on an Internet of Things approach using a NodeMCU ESP8266 microcontroller integrated with an MQ-135 gas sensor and a DHT11 temperature sensor. Environmental data are monitored in real time and displayed locally through an I2C LCD as well as remotely via the Blynk application. Experimental evaluation shows that the system is able to detect abnormal gas concentrations and temperature variations that indicate potential hazardous conditions. The results demonstrate that the proposed system can function as an effective early warning tool to support safety management and risk mitigation in chemical laboratory environments.

Halal Ingredient Detection in Packaged Food Products Using Multi-Layer Perceptron (MLP)

2026-01-20T13:43:04+07:00

The halal status of ingredients in packaged food products is a significant concern for Muslim consumers, yet variations in label formats and technical terminology often hinder manual verification. This study introduces Halal Ingredient Detection in Packaged Food Products Using a Multi-Layer Perceptron (MLP). A dataset of 55,149 ingredient entries from OpenFoodFacts was automatically labeled using internationally recognized lists of prohibited ingredients. Preprocessing included case folding, stopword removal, and TF-IDF text representation. Various MLP architectures were evaluated by considering macro F1-score, training time, and model generalization. The best-performing architecture was a simple MLP with 16–8 neurons, using the Adam optimizer and binarry cross-entropy loss. Using an 80:20 training–testing data split, the proposed MLP model achieved an accuracy of 94%, with the confusion matrix indicating low misclassification rates and strong discrimination between halal and non-halal ingredients. These results demonstrate that a straightforward MLP architecture combined with TF-IDF is sufficient to capture relevant textual patterns, providing an efficient and reliable approach for automated halal ingredient classification.

Application of the K-Medoid Algorithm to Cluster Percentage Data Based on Urban and Rural Areas in Indonesia

2026-01-20T16:34:34+07:00

This study applies the K-Medoids clustering algorithm to group Indonesian regions based on the percentage of ever-married women aged 15–49 years who gave birth in health facilities. The data used are secondary data obtained from Statistics Indonesia (BPS). The K-Medoids algorithm was chosen due to its robustness against outliers compared to K-Means [1]. The results show that regions can be grouped into clusters representing high and moderate utilization of health facilities for childbirth. This clustering can assist policymakers in identifying regional disparities and improving maternal health services.

Sentiment Analysis of Honkai Star Rail Game Reviews on Google Play Using the Naive Bayes Algorithm

2026-01-20T13:24:16+07:00

Honkai Star Rail is a turn-based JRPG developed by COGNOSPHERE PTE. LTD. that has attracted a large number of players with over 10 million downloads and 339,000 reviews on the Google Play Store. While most reviews are positive, some users have expressed their dissatisfaction. Sentiment analysis is crucial for extracting insights from reviews without having to read the entire text. The objective of this research is to conduct sentiment analysis on Honkai Star Rail users using the Naïve Bayes algorithm. This research employs the Naïve Bayes method due to its simplicity and ability to produce accurate predictions. Review data was collected through scraping from the Google Play Store and analyzed using the TF-IDF technique for word weighting. The model testing results showed the highest accuracy on test data using split validation with a 75:25 ratio, achieving 84.7%. The evaluation of the confusion matrix for positive sentiment resulted in a precision of 100%, recall of 12%, and an F1-Score of 22%, while for negative sentiment, it resulted in a precision of 84%, recall of 100%, and an F1-Score of 92%. This research contributes by providing feedback for game developers to improve quality and enhance player satisfaction

Application of Categorical Naive Bayes for Classifying Student Stress Levels Based on Academic and Lifestyle Factors

2026-01-20T22:07:14+07:00

Student stress has become an important issue in higher education due to increasing academic demands and lifestyle pressures. Early identification of student stress levels is necessary to support appropriate interventions. This study applies a machine learning approach to classify student stress levels based on academic and lifestyle factors using the Categorical Naive Bayes algorithm. The dataset was obtained from a student stress survey consisting of categorical attributes and stress level labels ranging from 1 to 5. Data preprocessing was conducted to ensure compatibility with the classification model, followed by data splitting into training and testing sets using an 80:20 ratio. Model performance was evaluated using a confusion matrix and standard classification metrics, including accuracy, precision, recall, and F1-score. The experimental results show that the proposed model achieved an accuracy of 0.50, with weighted average values of 0.52 for precision, 0.50 for recall, and 0.50 for F1-score. These results indicate that Categorical Naive Bayes is capable of performing stress level classification with moderate performance on categorical survey data. This study demonstrates the potential of machine learning techniques as a supporting tool for analyzing student stress levels and provides a basis for further model improvement in future research.

Development of a Web-Based MSME Digital Transformation Readiness Evaluation System Using a Design Science Research Approach

2026-01-21T08:08:59+07:00

Micro, Small, and Medium Enterprises (MSMEs) are still encountering challenges in gauging their overall preparedness for digital transformation. The goal of this study is to create an online platform that evaluates the digital readiness of MSMEs, offering a clear and organized assessment of their preparedness. The research approach employed is Design Science Research (DSR), which includes phases such as identifying the problem, defining the solution's goals, designing and creating the system, demonstrating its use, evaluating its effectiveness, and communicating and documenting the process. The outcome is an assessment tool that uses a survey to evaluate digital readiness across four key areas: technology, personnel, business workflows, and organizational strategy and culture. The results of the study reveal that the tool can effectively be applied by MSMEs and helps to discern variations in readiness across each area. It is anticipated that this system will assist MSMEs in establishing priorities and strategies for a more precise and sustainable approach to digital transformation.

Developing an Ethnomathematics Educational Game for Fractions Based on Local Wisdom (Case Study: SD Inpres Laipandak)

2026-01-21T17:06:14+07:00

Mathematics learning in elementary schools often faces obstacles, especially in conveying abstract concepts such as fractions. Students' low understanding is caused by the lack of connection between the material and everyday life. An ethnomathematics approach, which connects mathematical concepts with local culture, can be a solution. This study aims to develop an ethnomathematics-based educational game by integrating local wisdom from East Sumba in fraction learning. The method used in this study is Research and Development (R&D) with the ADDIE (Analysis, Design, Development, Implementation, and Evaluation) Development model. The game was developed by embracing local culture such as the distribution of meat in traditional ceremonies (kaparakang), ikat weaving patterns, and the division of agricultural land as learning contexts. A trial was conducted on elementary school students to assess the game's effectiveness in improving student understanding

Application of Support Vector Machine (SVM) Algorithm in Analyzing Public Sentiment in the Media Tiktok Related to Tour de Entete Event

2026-01-21T21:57:01+07:00

This study aims to analyze public sentiment on TikTok social media towards the implementation of the Tour De EnTeTe event using the Support Vector Machine (SVM) algorithm. The phenomenon of using social media as a space for public opinion provides an opportunity for local governments to evaluate the psychological and social impact of sports tourism activities in real-time. The research data was obtained through a crawling technique that produced 1,000 data entries, which were then reduced to 808 valid data after going through the text preprocessing stage. The analysis is done by classifying the data into positive and negative sentiments. The results showed that the SVM model was able to provide excellent performance with an accuracy rate of 89%, precision of 91%, and recall of 97%. The findings show a significant dominance of positive sentiment, reflecting the high enthusiasm and support of the public for the Tour De EnTeTe. Although there are constraints on language ambiguity that lead to some misclassification, overall the SVM algorithm has proven to be effective and accurate as an instrument for digital opinion analysis. This study concludes that the public perception of the event is very positive, so that it can be used as a strategic reference in the development of tourism promotion in East Nusa Tenggara in the future.

Implementation of Searching Algorithm for Optimizing Web-Based Product Identification Sales System

2026-01-20T22:08:21+07:00

The development of digital computer technology has brought major changes in the business world, especially in product promotion and sales. One of the rapidly growing businesses is distro (distribution store), a shop that sells local fashion products with unique designs and limited stock. Distro has become a means for young designers to express their creativity and meet the lifestyle needs of the community, especially teenagers and adults. However, the sales system of many distro stores is still done manually, starting from transactions, promotions through brochures and limited access for shop owners to recording financial reports of transactions. This causes various obstacles such as limited consumer reach, less updated product information, wasteful promotional costs and the risk of losing best-selling product data. To overcome these problems, the research aims to design and build an interactive web-based distro clothing sales system. This e-commerce website is designed to be accessible anytime and anywhere by consumers to see the latest product information, product categories and facilitate transactions and sales recording. With this system, it is hoped that it can expand the market, increase promotional efficiency and provide convenience and satisfaction for consumers in supporting business digitalization in the current technological era

Extreme Gradient Boosting for Daily Rainfall Forecasting in Medan City

2026-01-21T22:02:35+07:00

Curah hujan tinggi dan tidak teratur di Kota Medan sering menyebabkan banjir, sehingga peramalan curah hujan yang akurat sangat penting untuk mitigasi banjir. Studi ini bertujuan untuk meramalkan curah hujan harian dengan menerapkan metode Extreme Gradient Boosting (XGBoost). Proses pemodelan melibatkan pra-pemrosesan data, rekayasa fitur, dan standardisasi data. Kebaruan penelitian ini terletak pada integrasi XGBoost dengan teknik rekayasa fitur tingkat lanjut dan ambang batas kepentingan fitur untuk meningkatkan efisiensi dan akurasi model. Hasil evaluasi menunjukkan bahwa, dengan menggunakan rasio pembagian data 90:10, model mencapai MAE sebesar 1,05 dan RMSE sebesar 1,91. Analisis kepentingan fitur mengungkapkan bahwa CH_diff7, CH_diff1, dan CH_lag1 adalah prediktor yang paling dominan dalam peramalan curah hujan harian. Lebih lanjut, akurasi model ditingkatkan melalui pemilihan fitur, di mana penggunaan lima fitur teratas mengurangi MAE menjadi 0,99 dan RMSE menjadi 1,66. Temuan ini menunjukkan bahwa metode XGBoost, dikombinasikan dengan rekayasa fitur dan proses pemilihan fitur, memberikan pendekatan yang efektif untuk peramalan curah hujan harian di Kota Medan.

Implementation of Convolutional Neural Network for Rice Leaf Disease Classification to Optimize Farmers’ Decision-Making

2026-01-28T14:24:47+07:00

This study aims to develop a rice leaf disease classification model using a Convolutional Neural Network (CNN) with the MobileNetV2 architecture to assist farmers in making accurate decisions. A quantitative approach was employed through experimental methods involving the processing of 3,829 digital images from a publicly available dataset. The results indicate that the developed CNN model effectively classifies six categories of rice leaf conditions with 91% accuracy and was successfully integrated into a web-based application. This research concludes that the implementation of the MobileNetV2 architecture provides a rapid and efficient approach to plant disease diagnosis compared to traditional manual methods.

Classification of Obesity Risk Levels in Adolescents Based on Lifestyle using the Random Forest Algorithm

2026-01-23T07:51:24+07:00

Obesity in adolescents is a health problem influenced by diet, physical activity, and sedentary habits. This study aims at the classification of obesity risk in adolescents using a classification approach with the Random Forest algorithm based on the Obesity Levels and Lifestyle Dataset obtained from the Kaggle platform. The research stages include data preprocessing, dividing training and test data, model training, and performance evaluation using accuracy, precision, recall, and F1-score metrics. The test results showed an accuracy value of 93% with an average F1-score value of 0.93, indicating that the model has consistent performance in distinguishing various levels of obesity risk. Feature importance analysis shows that in addition to physical factors such as body weight, lifestyle factors such as physical activity and food consumption patterns also contribute significantly to the classification process. These findings confirm that the Random Forest algorithm is effective as a tool for early classification of obesity risk based on lifestyle, and can be the basis for developing a preventive adolescent health monitoring system.

An Android-Based Learning Multimedia for Intoducing Central Sumba Traditional Culture at SD Masehi Matawai Pandang

2026-01-23T07:50:52+07:00

Indigenous culture is a system of values and habits that is inherited from generation to generation and plays an important role in shaping cultural identity. However, there are still many students who do not know the traditional culture of their own region, as happened at SD Masehi Matawai Pandang, Central Sumba Regency. Students' low understanding of local culture is caused by the limitations of learning media that are able to increase students' interest in learning, because the learning process is still dominated by the use of package books without interactive media. This research aims to develop game-based interactive learning multimedia as a medium to introduce Central Sumba culture and customs for elementary school students. Game development uses the Game Development Life Cycle (GDLC) method, while the Linear Congruential Generator (LCG) algorithm is applied to generate pseudo-random numbers in the game. The results of the development are expected to increase students' interest in learning, students' understanding of indigenous culture, and foster a sense of belonging to local cultural heritage. In addition, this learning media is expected to be able to support digital cultural preservation and improve the quality of learning in remote areas

Development of an Android-Based Learning Media to Enhance Students’ Understanding of Microsoft Office Applications

2026-01-23T11:59:07+07:00

The rapid development of information technology has significantly influenced learning methods in secondary education. However, the learning process of Information and Communication Technology (ICT) subjects in several schools is still constrained by limited computer facilities and learning resources. This study aims to develop an Android-based learning media to enhance students’ understanding of Microsoft Office applications, namely Microsoft Word, Microsoft Excel, and Microsoft PowerPoint, for eighth-grade students at SMP Muhammadiyah Waingapu. The learning media was developed using the Multimedia Development Life Cycle (MDLC) method, which consists of concept, design, material collecting, assembly, testing, and distribution stages. The application was implemented using Unity and the C# programming language. To evaluate its effectiveness, a one-group pre-test and post-test design was employed involving 30 students. In addition, usability evaluation was conducted using the System Usability Scale (SUS). The results show a significant improvement in students’ learning outcomes, with the average score increasing from 60 in the pre-test to 90 in the post-test, yielding an N-Gain value of 0.75, categorized as high. Furthermore, the usability evaluation resulted in a SUS score of 91.67, indicating excellent usability. These findings demonstrate that the developed Android-based learning media is effective and suitable for supporting ICT learning, particularly in schools with limited computer facilities.

Design and Development of a Service Quality Measurement System Using the SERVQUAL Method : A Case Study of PT Karya Mandiri Engineering

2026-01-23T14:40:54+07:00

Service quality plays a critical role in shaping corporate image, operational sustainability, and customer satisfaction. This study examines service quality management at PT. Karya Mandiri Engineering, an engineering services and mechanical contracting company that has experienced notable fluctuations in operational activities over recent years, indicating instability in service performance. The study adopts the SERVQUAL method to measure gaps between customer expectations and perceptions across key service quality dimensions. The results show negative gap values in the dimensions of Reliability, Empathy, and Responsiveness, which are generally categorized as satisfactory. However, several service attributes remain in the fairly satisfied category, with one attribute showing a significantly large gap that indicates a priority area for improvement. These findings suggest that while certain service dimensions have met customer expectations, others require structured corrective actions. Based on the analysis, the study proposes the design of a web-based information system to monitor service performance and systematically evaluate service quality gaps. The proposed system is expected to support managerial decision-making, enhance service consistency, and contribute to more sustainable service quality management.

Web-Based Worship Schedule Information System for Gereja Jemaat Pamalar

2026-01-24T23:09:17+07:00

The scheduling of worship activities is a crucial aspect of church management, particularly for GKS Jemaat Pamalar, which conducts routine weekly and monthly activities. Manual scheduling often leads to various problems, such as time inaccuracies, data errors, and delays in delivering information to the congregation. Therefore, this study aims to design and develop a web-based worship scheduling system to assist church administrators in managing and disseminating worship schedules in a more organized and efficient manner. The system was developed using the Waterfall method, which consists of requirements analysis, system design, implementation, and testing stages. The main features of the system include worship schedule input, management of worship officers’ data, and direct publication of schedules through a responsive and user-friendly web interface that can be accessed on various devices. The implementation of this system is expected to improve the organization of worship activity management, reduce scheduling errors, and accelerate the dissemination of information to the congregation. The developed system functions as intended in supporting administrators in managing worship schedule data in a structured manner and presenting worship schedules clearly to the GKS Jemaat Pamalar congregation. Based on Black Box Testing, all system features were successfully tested with a 100% success rate.

Food Menu Recommendation System for Cholesterol and Diabetes Patients using Fuzzy Logic

2026-01-28T12:23:10+07:00

The number of people with cholesterol and diabetes continues to rise in modern society, mainly due to unhealthy diets. Many patients struggle to choose safe foods because nutrition levels vary. This research aims to develop an AI-based food recommendation system using Fuzzy Logic, which mimics human reasoning in handling uncertain data. The process involves fuzzification, inference, and defuzzification. By inputting data on sugar, cholesterol, fat, and calories, the system provides personalized and flexible food recommendations to help patients maintain a healthy diet.

Development of Educational Games to Improve Students’ Understanding of Basic Arithmetic Operations Based on Monte Carlo Algorithms

2026-01-27T08:40:19+07:00

This study aims to develop an educational game to improve the understanding of basic arithmetic operations for fourth-grade students of SDM Pahomba. Application development uses the Multimedia Development Lie Cycle (MDLC) method which consists of 6 phases: Concept, design, material collecting, assembly, testing, and distribution. In addition, the Monte Carlo Algorithm is applied to generate random questions to create a more interesting and adaptive learning process. The results of this study prove that the use of educational games is effective in improving students' understanding of basic arithmetic operations. This is evidenced by the success of black box testing which ensures all features function properly. And evidenced by a significant increase in the average score of 19 students, namely from a pre-test score of 45.25 to 84 during the post-test. This success is strengthened by the results of the calculation of the percentage of effectiveness of 85%. So the conclusion is that this educational game application is worthy of being used as a learning tool to improve students' understanding of basic arithmetic operations.

Stages of Facial Recognition Prediction using Artificial Intelligence Teachable Machine Learning Approach

2026-01-28T13:03:27+07:00

Artificial Intelligence (AI) has become vital part of everyday life and impacts various aspects, particularly how humans work and interact. One of the most widely developed applications of AI is facial recognition system integrated with the Internet of Things (IoT) and big data. This study discusses teachable artificial intelligence-based facial recognition prediction using a machine learning approach as the process reference, using two test scenarios: training images and testing using webcam, with accuracy as the benchmark. The data used comprises five age categories: 18–20, 21–30, 31–40, 41–50, and 51–60 years. Each category contains 25 images, resulting in total of 125 images for training and 50 images for testing. The results show that highest accuracy rate is found in the 18–20 age category, with 89% accuracy. Other age categories exhibit lower accuracy variations and experience classification errors. In the 51–60 age group, the model achieved 66% accuracy with 50x epoch settings, batch size 16, and learning rate 0.001. Webcam testing highest accuracy of 86% in the 21–30 age group. These results demonstrate that teachable machines can be used as an initial experimental tool in AI model development before implementation in software or hardware.

Application of Machine Learning in Predicting Parking Duration Categories to Support Campus Land Occupancy Management

2026-01-30T20:30:50+07:00

The uncertainty of parking space availability during peak hours is a major obstacle in operational management at Esa Unggul University, Bekasi Campus. Abundant parking transaction data has so far only been stored as administrative archives without being utilized for strategic decision-making. This study aims to build a prediction model for parking duration categories (Short-Term vs. Long-Term) to support proactive occupancy management. Using a dataset of 1,329 valid transaction records, this study compares the performance of Random Forest and Naive Bayes algorithms. Experimental results show that Random Forest is superior in capturing parking behavior patterns with an accuracy reaching 72.18% and a sensitivity level (Recall) of 79.4% in the Long-Term class. Data pattern findings indicate that vehicles entering between 07:00–09:00 WIB have a dominant probability of parking for a long duration. As a managerial implication, the model's prediction results are integrated into a Web-based Dashboard Monitoring prototype that presents occupancy visualization and traffic trends in real-time. This system is expected to assist parking management in implementing more efficient traffic diversion and parking slot allocation strategies

Utilization of MATLAB as a Learning Medium in Basic Mathematical Computation Techniques

2026-01-30T20:33:53+07:00

Mathematical computational learning is an important component in mastering numerical and programming concepts. MATLAB as a numerical computation software provides various facilities to support the learning process, especially in basic computational engineering materials such as finite iteration, conditioned iteration, and logic processing. This study aims to analyze the use of MATLAB as a medium for learning basic mathematical computational techniques through a series of practicums that include the use of if-else control structures, for loops, while loops, and switch cases. The method used is a practicum experiment by running several MATLAB programs to solve mathematical problems, such as calculating final values, factorials, finite iterations, and calculating the volume of geometric shapes. The practicum results show that MATLAB is able to visualize the computational process effectively, provide direct feedback, and strengthen students' understanding of mathematical logic and algorithm concepts. Thus, MATLAB can be an efficient learning medium in improving students' basic computational skills in the field of numerical methods.

Analysis of Critical Success Factors in Information Technology Project Management: A Literature Review

2026-01-29T09:31:56+07:00

This study examines the factors influencing the success of information technology (IT) project management through a literature study approach. The high failure rate of IT projects, characterized by schedule delays, cost overruns, and unmet objectives, highlights the importance of effective project management practices. The objective of this research is to analyze and synthesize key factors that contribute to the success of IT project management based on existing academic literature. The research method employed is a literature study, involving the review and analysis of relevant national and international journals, textbooks, and scholarly publications related to IT project management. The findings indicate that several critical factors consistently influence IT project success, including comprehensive project planning, competent human resources, effective risk management, clear communication among project stakeholders, and the use of project management support tools. These factors are interrelated and collectively determine the ability of organizations to manage IT projects efficiently and effectively. The implications of this study provide a conceptual framework that can be used by practitioners to improve IT project performance and by academics as a reference for future empirical research in the field of information technology project management.

Analysis of Public Perception of the Digital Branding of the “Content Governor” on X Social Media using Naïve Bayes

2026-01-29T09:31:21+07:00

The development of social media, particularly the X (Twitter) platform, has created a digital discussion space that influences the formation of public opinion on political issues. The “Content Governor” phenomenon, often associated with Dedi Mulyadi, has generated various responses from users, making systematic sentiment analysis necessary to understand these response patterns. This study aims to map public sentiment tendencies while developing a sentiment-classification model using a Naïve Bayes approach combined with TF–IDF weighting. Data were collected through the Twitter API based on several relevant keywords, yielding 2,133 tweets, which, after a selection process and manual labeling, were narrowed down to 1,023 labeled data points consisting of positive, neutral, and negative sentiments. The research stages include preprocessing (case folding, cleaning, normalization, tokenization, stopword removal, and stemming), data balancing using SMOTE, TF–IDF feature construction, and training a Multinomial Naïve Bayes model with an 80:20 data split ratio. Evaluation was carried out using accuracy, precision, recall, F1-score, and a confusion matrix. The results show that the model achieved an accuracy level of 74.15%, with the strongest performance in the negative sentiment category. The distribution of public sentiment also indicates a dominance of negative sentiment (812 items) compared to neutral and positive sentiments, suggesting that netizens’ responses to the “Content Governor” issue tend to be more critical and negative in tone. These findings are expected to enrich studies on digital political communication and provide methodological contributions to sentiment analysis research in the Indonesian language.

Sentiment Analysis of Public Opinion on the 2024 President-Elect's Administration on Twitter using Naïve Bayes

2026-01-29T21:08:48+07:00

The increased use of social media, especially Twitter, has created a need for systematic analysis to understand public opinion on political issues, including the performance of the president-elect in 2024. This study analyzes public opinion on these issues using the Naïve Bayes algorithm. Data was collected using scraping techniques and then divided into three sentiment categories positive, negative, and neutral. After the labeling process, the data underwent preprocessing, which included data cleaning, case folding, normalization, tokenization, stop word removal, and stemming. TF-IDF weighting was used to represent features, while the SMOTE technique was applied to balance class distribution. A total of 1,074 tweets were analyzed. The results showed that negative opinions dominated at 59.9%, followed by positive opinions at 29.8% and neutral opinions at 10.3%. Model performance evaluation showed that Naïve Bayes was able to consistently identify sentiment patterns, with 71% accuracy, 74% precision, 71% recall, and an F1 score of 72%. These results prove that the combination of TF-IDF and SMOTE contributes significantly to improving classification effectiveness. This study provides a comprehensive overview of trends in public opinion.

Improving the Accuracy of Early Diagnosis of Dengue Hemorrhagic Fever Based on Clinical Symptoms Using Random Forest

2026-01-29T21:16:52+07:00

The development of machine learning in the field of health provides important opportunities for improving the accuracy of disease diagnosis, including dengue hemorrhagic fever (DHF), which remains a major health problem in Indonesia. This study aims to develop an early DHF diagnosis model based on the Random Forest algorithm using clinical symptom data from patients at the Kosasih Group Clinic. The research was conducted using a quantitative approach through the CRISP-DM stages, which included data acquisition, validation, cleaning, and preprocessing, covering missing value handling, normalization, and class imbalance management using SMOTE. The dataset was then divided using stratified sampling to maintain class proportions, followed by training the Random Forest model optimized using Bayesian Optimization to obtain the best combination of hyperparameters. Performance evaluation was carried out using accuracy, precision, recall, F1-score, and ROC-AUC metrics, and validated using stratified k-fold cross-validation to ensure model stability. Model interpretability was analyzed using SHAP and LIME to identify the contribution of each clinical symptom to the prediction. The results showed that the model was able to provide high classification performance, with increased sensitivity to DHF cases after applying SMOTE and consistent interpretation of clinical symptoms such as fever, joint pain, and nausea. These findings confirm the potential of Random Forest as a reliable model to support the development of AI-based clinical decision support systems (CDSS) for early diagnosis of DHF in primary health care facilities.

Development of a Role-Based Mobile Application for Operational Digitalization in Sumatra Jewelry Store

2026-02-05T20:21:20+07:00

Operational activities in many jewelry stores are still managed manually using paper-based records, which often cause miscommunication, data loss, and delays in order processing. This study focuses on the development of a role based mobile application to support the digitalization of operational processes in a jewelry store environment. The application was developed using the Waterfall method and implements a client-server architecture with a relational database to manage orders, task assignments, and inventory data in an integrated manner. Each user role, including sales, workers, and inventory administrators, is provided with specific access rights and functionalities according to their responsibilities.. The results show that the proposed application improves workflow transparency, enhances data accuracy, and simplifies real-time order tracking across different operational stages. The digital system also assists store owners in monitoring performance and making more informed decisions. Overall, the implementation of the mobile application demonstrates its effectiveness in improving operational efficiency and reducing miscommunication compared to the previous manual system.

Design and Build a Web-Based Attendance Information System with Location Validation in the Nasdem Fraction

2026-01-26T11:17:44+07:00

Manajemen kehadiran karyawan memiliki peran penting dalam mendukung disiplin dan efektivitas administrasi organisasi. Namun, proses kehadiran yang masih dilakukan secara konvensional berpotensi menimbulkan masalah, seperti data yang tidak akurat, keterlambatan pencatatan, dan validasi kehadiran yang lemah. Studi ini bertujuan untuk merancang dan membangun sistem informasi kehadiran karyawan berbasis web dengan validasi lokasi di Faksi Nasdem untuk meningkatkan akurasi dan efisiensi pencatatan kehadiran. Metode pengembangan sistem yang digunakan adalah Siklus Hidup Pengembangan Perangkat Lunak (SDLC) dengan model Waterfall, yang meliputi tahapan analisis kebutuhan, desain sistem, implementasi, pengujian, dan pemeliharaan. Sistem dikembangkan menggunakan bahasa pemrograman PHP dan basis data MySQL serta dilengkapi dengan fitur validasi lokasi berbasis GPS dan pengambilan foto sebagai bukti kehadiran. Pengujian sistem dilakukan menggunakan metode Black Box Testing. Hasil penelitian menunjukkan bahwa sistem yang dikembangkan mampu mencatat kehadiran dan keluar secara digital, memvalidasi lokasi kehadiran, dan menyajikan laporan kehadiran secara terstruktur. Sistem ini dianggap sesuai untuk digunakan dan berkontribusi pada peningkatan efisiensi administrasi dan keandalan data kehadiran karyawan.

Expert System for Diagnosing Diseases in Corn Plants using Forward Chaining and Certainty Methods Web-Based Factor

2026-01-29T18:27:21+07:00

Corn is one of the main agricultural communities that plays an important role in supporting community food security. In Ndapayami Village, East Sumba Regency, corn farming is the main source of livelihood for the community. However, the productivity of corn crops often decreases due to disease attacks that are difficult for farmers to identify quickly. This study aims to design and build an expert system in diagnosing diseases in corn plants using the Forward Chaining method as an inference engine and Certainty Factor to measure the level of certainty of diagnosis. The expert system is designed to be computer-based so that it is easily accessible to farmers and agricultural extension workers in determining solutions to handle corn crop diseases. The system was developed using a rule-based approach and tested through the black-box testing method in 15 case scenarios. The research data was obtained through field observations, interviews, and documentation, and validated by agricultural experts. The test results show that the expert system has an accuracy rate of 80% and is able to give a confidence value to each diagnosis result. This system is expected to help speed up the disease diagnosis process, increase farmers' knowledge, and support increased productivity and food security in Ndapayami Village.

Scheduling Information System Design Biblical Understanding (PA) In GKS Jemaat Wulla Website-Based

2026-01-29T21:24:17+07:00

The development of information technology in the digital era has a significant influence on various areas of life, including religious services. The Sumba Christian Church (GKS) Jemaat Wulla t is one of the congregations that routinely carries out PA activities at the household, environment, and other service categories. However, the PA scheduling process that is still carried out manually causes various problems, such as overlapping schedules, and delays in obtaining information on the schedule of the current understanding. This research aims to design and build a website-based PA scheduling information system to assist the GKS Jemaat Wulla in managing PA activities more effectively, efficiently, and structured. The development of the system is carried out using the Extreme Programming (XP) method, an Agile approach that emphasizes flexibility, intense collaboration with users, and continuous testing to ensure that the resulting software is of good quality and adaptive to changing needs. The XP stages that are implemented include planning, designing, coding, and testing. The results of this study resulted in a website-based PA scheduling information system that is able to automate the scheduling process, reduce the potential for manual errors, and provide real-time access to information to administrators and congregations. This system is expected to support increasing the effectiveness of church services, facilitate coordination of PA implementation, and encourage jemaat participation through the provision of more accurate, fast, and transparent schedule information.

Application of Artificial Neural Networks in Predicting the Number of Out-of-School Children Based on Expenditure Groups in Indonesia (2019–2023)

2026-01-29T21:26:11+07:00

Out-of-School Children (OSC) is one of the key indicators used to evaluate the equity of access to education in Indonesia. Data from Statistics Indonesia (BPS) indicate that OSC rates are influenced by differences in household expenditure groups across primary school (SD), junior high school (SMP), and senior high school (SMA) levels. This study aims to apply an Artificial Neural Network (ANN) method to predict OSC values based on historical patterns of household expenditure data. The data used in this study are secondary data from BPS covering the period 2019–2023, which are grouped into five expenditure quintiles. The ANN model employed is a Multilayer Perceptron consisting of four input neurons, two hidden layers with 24 neurons each, and one output neuron. The training process is conducted using the backpropagation algorithm with a hyperbolic tangent activation function and Min-Max Scaling for data normalization. The results indicate that the ANN model is able to consistently learn the relationship patterns between household expenditure groups and out-of-school children rates. The trained model is further used to simulate predictions of OSC values for subsequent periods. This study is expected to serve as an alternative computational approach for analyzing education indicators based on socio-economic data.

Development of an Android-Based Educational Game Using Voice Recognition for German Vocabulary Pronunciation Training

2026-01-30T09:47:53+07:00

German language learning at SMA Negeri 1 Rindi Umalulu faces several challenges. Teaching methods remain conventional, focusing on reading and writing vocabulary from textbooks followed by group repetition, with minimal individual practice. As a result, many students lack confidence in pronouncing vocabulary independently due to fear of making errors. Supporting facilities such as language laboratories, audio-visual media, and interactive applications are very limited, making it difficult for teachers to provide accurate pronunciation models. Consequently, students often imitate vocabulary with incorrect intonation. In addition, learning motivation is low because vocabulary learning is seen as memorization rather than understanding proper pronunciation. This condition leads to weak speaking skills, especially in pronouncing basic German vocabulary such as greetings. This study used the Multimedia Development Life Cycle (MDLC) method to develop an interactive learning medium. The Fisher–Yates Shuffle algorithm was applied to randomize questions, and voice recognition technology was used to train pronunciation. The results showed an improvement in students’ pronunciation ability, with the average score increasing from 62.18 to 74.37. Black Box Testing achieved a 100% success rate, and the System Usability Scale (SUS) score reached 82.7, categorized as excellent.

Transformation of Vocational Education Assessment Architecture Integration of Digital Portfolio-Based E-Report Card System and Industry 4.0 Competencies in the Communication Study Program

2026-02-04T10:27:29+07:00

In the rapidly evolving vocational education ecosystem under the shadow of the Industrial Revolution 4.0, there is a fundamental disconnect between conventional scholastic assessment methods and the dynamic competency demands of the creative industry. This research report presents an in-depth investigation into the design and validation of an academic information system architecture (e-Rapor) expanded, specifically calibrated to the needs of the Communication Study Program (Multimedia, Visual Communication Design, and Broadcasting). Against the backdrop of the high Open Unemployment Rate (TPT) of Vocational High School (SMK) graduates in Indonesia, which reached 8.62% in 2024, this study identified the failure of the standard numerical assessment system in capturing the essence of students' productive competencies and creativity. Adopting the methodology Design Science Research(DSR) which is synergized with development techniques Rapid Application Development(RAD), this research has successfully designed a hybrid assessment ecosystem. This system functions not only as an administrative assessment repository but also as a talent management platform that integrates a digital portfolio based on cloud, project-based authentic assessment (Project-Based Learning), and competency validation aligned with the Indonesian National Work Competency Standards (SKKNI). A case study was conducted in the strategic context of the Karawang industrial area, West Java, involving a network of local vocational schools and universities to test the system's relevance to the needs of the international industrial area. The research findings indicate that the integration of multimedia artifact curation features, dynamic assessment rubrics, and industry feedback mechanisms significantly improves the validity of the measurement. hard skills And soft skills(4C), as well as providing more accurate competency verification instruments to bridge the gap between supply vocational graduates and demand professional workforce

Simple Linear Regression Models for Peanut Production Prediction in North Sumatra 2022

2026-01-31T09:50:22+07:00

Peanuts (Arachis hypogaea L.) are a palm oil commodity that functions strategically in supporting food security and the regional economy. North Sumatra Province has great potential in peanut development, but increasing production is still faced with a number of obstacles. This study aims to examine the relationship between harvest area and peanut production as well as assess the ability of harvest area as a predictor variable of production in North Sumatra Province. This study applies a quantitative approach using a simple linear regression analysis method. The data analyzed is secondary data obtained from the Central Statistics Agency (BPS) of North Sumatra Province through the publication of North Sumatra in Figures 2022, which contains data on harvest area and peanut production at the district/city level. Of the 27 districts/cities, as many as 26 districts/cities were analyzed after going through the data filtering process.

The results of data processing produced a regression equation Y=12.356+0.000592X, which showed that the effect of harvest area on peanut production was positive but very weak. A determination coefficient value (R²) of 3.09% indicates that the harvest area is only able to explain a small part of the variation in production, while most of it is influenced by other factors outside the model. Significance testing through the F-test and t-test showed that the regression model was not significant at a 95% confidence level. These findings confirm that increasing peanut production cannot rely solely on land expansion, but rather requires an intensification strategy through increased productivity and input efficiency.

Plant Leaf Disease Classification Using Convolutional Neural Network Based on Digital Images

2026-02-01T08:16:45+07:00

Monitoring plant health is an important factor in maintaining agricultural productivity. Manual identification of leaf diseases requires expert knowledge and is prone to errors due to visual similarities among disease symptoms. This study aims to develop a plant leaf disease classification system based on digital images using a Convolutional Neural Network (CNN) approach. The dataset consists of plant leaf images representing three disease classes: Corn–Common rust, Potato–Early blight, and Tomato–Bacterial spot. Prior to model training, the images undergo preprocessing steps including image resizing and pixel normalization. The performance of the CNN model is evaluated using a testing dataset that is not involved in the training process, employing accuracy, confusion matrix, precision, recall, and F1-score as evaluation metrics. Experimental results show that the proposed model achieves a test accuracy of 95.56%, with balanced performance across all disease classes. In addition to quantitative evaluation, the trained model is implemented in a Streamlit-based application, allowing users to upload plant leaf images and obtain disease classification results interactively. The findings indicate that the CNN-based approach is effective for plant leaf disease classification and has potential application as an early decision-support system for plant health monitoring.

Development of Canva-Based Interactive Digital Learning Media with Hyperlink Features for Ecosystem Learning in Elementary School

2026-02-01T08:21:23+07:00

This study aimed to develop a digital interactive learning media based on Canva with hyperlink features for the science ecosystem topic in fifth-grade elementary school. The study employed a Research and Development approach using the 4D model, consisting of define, design, develop, and disseminate stages. The developed media integrated text, images, instructional videos linked to YouTube, and interactive assessments generated through Canva AI. Data were collected through expert validation sheets and practicality questionnaires administered to teachers and students. The results indicated that the developed learning media achieved a very good level of validity in terms of media design, content accuracy, and language use. Furthermore, the practicality test results showed very positive responses from both students and teachers, indicating that the media was easy to use, visually appealing, and engaging in the learning process. Based on these findings, the digital interactive learning media based on Canva hyperlink was concluded to be valid and practical for use as a supporting learning resource in elementary science learning on ecosystem topics.

Implementation of Prototyping Method in Lontara Studio Management Information System Bone Regency

2026-02-01T09:36:05+07:00

Art studios play an important role in preserving traditional dance as a valuable cultural heritage; however, many still face operational challenges such as limited promotion, manual administrative processes, and unstructured rental services that reduce management effectiveness and public access. This study aims to design a web-based Management Information System (MIS) for Sanggar Lontara in Bone Regency using the prototyping method to support administrative management and service rental activities in accordance with user needs. The system was developed through an iterative prototyping approach involving requirement analysis, design, implementation, evaluation, and refinement with active participation from studio administrators. System evaluation consisted of alpha testing using black-box testing and beta testing through a Likert-scale questionnaire distributed to 32 respondents across six assessment indicators. The developed system includes service catalogs, online booking, payment proof uploads, administrative verification, reporting, portfolio galleries, studio information, and role-based access for administrators, renters, and managers. Alpha testing demonstrated a 100% success rate for core functionalities, while beta testing achieved a user satisfaction index of 88.02%, classified as “Very Good,” indicating that the system is feasible for implementation with minor usability improvements.

Implementation of the Prototyping Method for a Web-Based BUMDes Appakabaji Management Information System in Maros Regency

2026-02-01T20:52:51+07:00

Operational management at the SPAM business unit of BUMDes Appakabaji is still carried out manually through visit records in books and has not yet been integrated into a digital system. As a result, data is difficult to trace and team work area distribution is not well organized. The growth in the number of customers, reaching 1,021 records, indicates the need for a more efficient management system. This study aims to develop a web-based management information system to improve the efficiency of customer data management, visit recording, work area mapping, payment management, and recapitulation reporting at BUMDes Appakabaji. The research employed the prototyping method, consisting of needs analysis, design, implementation, and testing stages. System testing using black box testing shows that the developed system functions properly and is feasible to improve operational efficiency.

Data Warehouse Design on Student Data, Faculty of Computer Science, UKIM

2026-02-02T10:07:18+07:00

The development of science and information technology cannot be separated from the role of universities as one of the institutions of secondary education. A university, especially within the scope of a university, certainly has a diverse number of students. Moreover, if the university has good quality of education and teaching quality, it will have an attraction for prospective students who want to continue their studies at university. The high number of students in a university, especially certain faculties, often results in hampered in the process of managing and analyzing student data. This research aims to design a Data Warehouse on student data of the Faculty of Computer Science, UKIM. This is done to facilitate the process of managing and analyzing student data of the Faculty of Computer Science. The Waterfall method is used in this study as a reference in designing a Data Warehouse. The results of this research can help the Faculty of Computer Science in terms of managing and analyzing student data.

Web-Based Internship Absence and Activity Information System Using GPS (Case Study: BMKG Region I, Medan City)

2026-02-06T16:36:40+07:00

Internship attendance and activity recording are crucial components of the human resource evaluation and development process. At the BMKG Region I, Medan City, the attendance and internship activity reporting process was previously carried out manually, potentially leading to recording errors, reporting delays, and difficulties in evaluating intern performance. This study aims to implement a web-based Internship Attendance and Activity Information System utilizing Global Positioning System (GPS) technology. The research method used was the System Development Life Cycle (SDLC) with a case study approach. The system was developed using the native PHP programming language with JSON-based data storage. The results showed that the system is capable of supporting location-based attendance, daily activity recording, activity verification by supervisors, and systematic intern performance assessment. The implementation of this system has been proven to increase the effectiveness and efficiency of internship management within BMKG Region I, Medan City.

A Systems Engineering Process Approach for AIoT-Based Export Supply Chain ESG Measurement

2026-02-05T14:20:22+07:00

The supply chain for export commodities is facing increasing demand driven by transparency and accountability regarding environmental, Social, and Governance (ESG) performance. However, ESG measurement in practice is still hampered by data distribution, reliance on manual reporting, and the absence of a framework for an engineering-structured system. Research: This aim is to design a system chain for ESG measurement of supply-export commodities and natural resources using a systematic Systems Engineering Process (SEP) approach. SEP is applied to transform raw data sourced from the chain sensor supply export into relevant ESG information for decision-making. Methodology study covering requirements analysis, system design, implementation, testing, deployment, and maintenance stages, to ensure integration between the operational needs and the technical solutions. Research results show that the SEP approach can provide a consistent, integrated framework that improves process traceability and supports the sustainability system in the long term. Approach. This expectation can serve as a base development system that is more reliable, adaptive, and applicable to the supply chain for export commodities.

Comparison of Decision Tree and Random Forest in Book Loan Classification for Universitas Esa Unggul Bekasi

2026-02-08T09:02:05+07:00

This study compares the performance of Decision Tree and Random Forest algorithms in classifying the status of book loans at the Library of Universitas Esa Unggul Bekasi Campus. The objective of this research is to build a predictive model capable of identifying potential late returns as a basis for more proactive decision-making. The dataset used consists of 1,210 historical book loan records from the period of January to May 2025. Preprocessing stages included data cleaning, feature engineering, encoding of categorical variables, and handling class imbalance using the Synthetic Minority Over-sampling Technique (SMOTE). Classification models were built and evaluated using accuracy, precision, recall, F1-score, and AUC-ROC metrics. Test results showed that the Random Forest algorithm had superior and more stable performance compared to Decision Tree, especially in detecting the minority class of late loans. After hyperparameter tuning, Random Forest achieved higher F1-score and recall values without a significant drop in precision. These findings indicate that Random Forest is more effective for handling imbalanced and complex loan data. Therefore, the Random Forest algorithm is recommended as a decision support system to improve service efficiency, collection availability, and library management quality.

Analysis of Internet Network Quality of Service (QoS) at Yamaha Central Office Palembang using Wireshark

2026-02-07T23:23:19+07:00

The rapid growth of internet usage in office environments demands reliable and consistent network performance to support daily operational activities. However, unstable network quality may lead to communication delays, data transmission errors, and decreased work productivity. Therefore, the research problem addressed in this study is the lack of objective evaluation regarding the actual Quality of Service (QoS) performance of office internet networks. This study aims to analyze and evaluate the Quality of Service of the internet network at the Yamaha Central Office Palembang by measuring key QoS parameters, including throughput, packet loss, delay, and jitter. The objective is to determine whether the existing network performance meets acceptable service quality standards for office operations. A quantitative descriptive method was employed in this research. Data were collected through direct observation and real-time packet capturing using Wireshark during active network usage within a specific measurement period. The captured network traffic data were then processed and analyzed quantitatively based on standard QoS measurement criteria. The results show that the network achieved a throughput of 2416 kbit/s, packet loss of 0.0%, an average delay of 3.53 ms, and jitter of 3.22 ms. These values indicate that the internet network performance at the Yamaha Central Office Palembang is classified as very good and complies with acceptable QoS standards. The contribution of this study lies in providing an empirical evaluation of office network performance using real traffic data and offering practical insights that can be utilized as a reference for network monitoring, performance optimization, and future QoS-related studies in similar office environments.

Mitigating Data Sparsity in Code-Mixed Text through Back-Translation Augmentation for Aspect-Based Sentiment Analysis in Tokopedia Reviews

2026-02-09T13:51:49+07:00

Aspect-Based Sentiment Analysis (ABSA) in e-commerce reviews in Indonesia faces significant challenges, including the use of mixed language or code-mixed language and limited labeled data, or data sparsity. This study proposes the use of Back-Translation data augmentation techniques to enrich Tokopedia's mixed Indonesian-English or South Jakarta language review dataset. Using the IndoBERT model, experimental results show a 3% increase in accuracy for both aspect and sentiment classification. These findings demonstrate that artificial data augmentation is effective in addressing data sparsity constraints in informal texts and improving the reliability of macro analysis for strategic platform recommendations.

Analysis and Design of Web-Based Laundry Management Information System at ELS Laundry using the RAD Method

2026-02-11T16:28:44+07:00

This significant technological advancement has affected almost all sectors of life, including the business world. ELS Laundry is a laundry service provider. However, in its operations, data management is still done manually. When customers come to drop off their items, data recording and note-taking are done manually, which are then recorded in a customer data ledger. This creates the potential for customer data loss or damage due to lost or torn paper. The manual and suboptimal transaction system causes problems such as the accumulation of archives and reports, slow information flow, and difficulties in searching for data and summarizing transaction reports. To overcome this problem, a Laundry Management Information System is needed to improve the effectiveness and efficiency of ELS Laundry's operations. This study aims to produce a web-based application using the PHP programming language and MySQL database by implementing the RAD development method to design a Laundry Management System. The result of this study is the development of a web-based application that can be used as an information medium to facilitate customer data management at ELS Laundry.

Application of Augmented Reality and First-Person Games in Introducing Gravity as a Learning Method in Astronomy

2026-02-12T20:02:31+07:00

The development of digital technology brings significant opportunities for learning innovation, particularly in the fields of science and astronomy. One technology that can be utilized is Augmented Reality (AR) combined with a first-person game-based approach. This research aims to develop interactive learning media that can help students understand the concept of gravity more easily and enjoyably. Through the application of AR, students can see the gravity values on each planet in three dimensions as if they were present in a real environment, while the first-person mode allows them to experience the experience of being directly in a space simulation. The implementation results show that this approach can increase the appeal of learning, motivate students to be more active, and strengthen conceptual understanding of gravity. Thus, the application of AR and first-person games in the introduction of gravity can be an innovative alternative to enrich science learning methods in the digital era.

Representation of Social Support in Social Media: Big Data Analysis to Understand the Cultivation of Religious and Ethical Character Values in the Digital Era

2026-02-13T13:33:11+07:00

The digital transformation in the Society 5.0 era has altered the landscape of character education, wherein conflicts between teachers, students, and bureaucratic systems are now openly exposed in the digital public sphere. This study aims to deconstruct public perception and moral responses regarding the issue of teacher criminalization through a Computational Sociology approach. Utilizing Big Data Analytics methods, this research analyzes 2,803 digital interactions (comments) on the YouTube platform related to conflict cases involving honorary teachers.

The analysis was conducted using Deep Learning algorithms (IndoBERT) for sentiment classification and Latent Dirichlet Allocation (LDA) for topic modeling, with a model validation rate reaching 75%. The results indicate a phenomenon of digital ethical paradox, where 70.3% of public responses were dominated by negative sentiments. However, topic analysis reveals that this negativity is not a form of hatred toward the teaching profession, but rather a manifestation of "Aggressive Social Support." Public outrage is polarized around three crucial issues: resistance to illegal levy practices, criticism regarding the degradation of student/parental etiquette (adab), and demands for school leadership accountability.

This research concludes that society's digital footprint is an authentic reflection of collective unrest regarding structural injustices within education. The implications of this study emphasize the need for a redefinition of character education that not only focuses on students but also encompasses digital ethical literacy for the school ecosystem and legal protection reform for the teaching profession.

Implementation of the Website-Based Prototyping Method for Marketing Management Information Systems for MSME Products in Pangkep Regency

2026-02-13T22:17:20+07:00

This research is motivated by the importance of website-based marketing digitalization to improve the competitiveness of MSMEs in Pangkep Regency, which still face limitations in integrated marketing systems, unstructured product catalog management, and suboptimal online transaction processes. This study aims to develop a website-based marketing management information system for MSME products in Pangkep Regency using the prototyping method and to evaluate its feasibility in terms of functionality, interface design, and ease of use. The research applied an iterative prototyping method, including requirements gathering, initial prototype development, user evaluation, and repeated refinement based on feedback from MSME actors and prospective users. The system was implemented using PHP and MySQL with security measures such as input validation, CSRF tokens, and data sanitization. System testing was conducted through alpha testing for functionality and beta testing using a Likert-scale questionnaire. The results indicate that the developed system successfully meets user needs with functional features and achieves an average score of 4.4 or an index of 89.03%, which falls into the “Excellent” category, showing that the system is highly suitable for use by MSMEs in Pangkep Regency.

Web-Based Inventory Management Information System for Catholic Church

2026-02-13T22:17:56+07:00

The rapid development of information technology encourages organizations, including religious institutions, to implement information systems that support operational activities. Inventory management at the Catholic Church of St. John the Baptist Perawang was previously conducted manually, resulting in recording errors, data duplication, and delays in report preparation. This study aims to design and develop a web-based inventory management information system using the Waterfall software development method. The development stages include requirement analysis, system design using Unified Modeling Language (UML), implementation using PHP and MySQL, and functional testing through Black-Box testing. The resulting system is capable of centrally managing inventory data, handling borrowing, returning, transfer, damage, and repair transactions, and generating structured reports automatically. The implementation results indicate that the system improves efficiency, data accuracy, and transparency in inventory management. The system provides an effective solution for supporting asset management and operational services within the church environment.

Design and Implementation of an Indigenous Knowledge Management Framework for The Dalihan Na Tolu System in The Batak Toba Community (Case of North Sumatra, Indonesia)

2026-02-15T21:57:29+07:00

This study presents the design and implementation of an Indigenous Knowledge Management Framework (IKMF) for the Dalihan Na Tolu system in the Batak Toba community of North Sumatra, Indonesia. Dalihan Na Tolu, consisting of Hula-hula, Dongan Tubu, and Boru, governs kinship relations, ceremonial obligations, inheritance systems, and customary conflict resolution. Despite its continued relevance, knowledge transmission remains predominantly oral and tacit. This research applies the SECI knowledge conversion model integrated with rule-based knowledge representation to formalize and digitize Dalihan Na Tolu knowledge. A mixed-method research design was employed involving interviews with traditional leaders (Raja Adat), elders, and surveys of community members. A prototype knowledge-based system was developed to demonstrate framework applicability. Evaluation using ISO 25010 quality attributes indicates strong usability and functional performance. The proposed framework contributes to indigenous knowledge preservation, digital cultural heritage, and sustainable governance models.

Classification of Weather Conditions in Medan City: Application of the C4.5 Decision Tree Algorithm as a Prediction Accuracy Solution

2026-02-16T21:06:25+07:00

The Accurate weather predictions play an important role in supporting community activities and economic sectors, especially in cities with varied weather patterns like Medan. Erratic weather can have a significant impact on the transportation sector, agriculture and daily community activities. This research aims to develop a weather condition classification model using the Decision Tree C4.5 algorithm, which is known to be reliable in handling numerical and categorical data. The historical data used includes temperature, humidity, wind speed and rainfall, which are analyzed using the CRISP-DM approach, including the stages of business understanding, data preparation, modeling and evaluation. This model classifies weather into four categories: sunny, cloudy, drizzly, and rainy. Based on test results, the model succeeded in achieving an accuracy level of 100%, showing its reliability in predicting the weather in Medan. Apart from providing accurate weather information, this model is also able to support data-based decision making, such as transportation planning and risk mitigation in the agricultural sector. These findings show the great potential of using the C4.5 algorithm in analyzing and predicting weather in Indonesia, as well as being a practical solution for facing complex weather challenges. This research provides an important contribution in supporting the development of data-based weather prediction technology that can be implemented widely for the benefit of society and the industrial sector

Comparative Analysis of Apriori, FP-Growth, and Eclat Algorithms in Determining Tourist Visit Patterns in Simalungun Regency

2026-02-19T15:59:40+07:00

Simalungun Regency possesses diverse tourism potential, yet its utilization remains suboptimal due to limited understanding of tourist visit patterns. This study aims to analyze visit patterns to tourist attractions in Simalungun Regency using three Association Rule Mining algorithms—Apriori, FP-Growth, and Eclat—implemented in a web-based system. The web system was designed using the Flask framework with a MySQL database to manage tourist visit data in real-time. The interface allows users to input visit data and perform interactive comparative analysis of the three algorithms with adjustable parameters. Results demonstrate that all three algorithms successfully identified strong visitation patterns, such as Lake Toba → Sipiso-piso Waterfall (confidence 82%, lift 1.85). In terms of performance, FP-Growth exhibited the fastest execution time (2.3 seconds at 10% support), followed by Eclat (4.8 seconds) and Apriori (12.5 seconds). Eclat proved most efficient in memory usage (85 MB). The developed web system facilitates the Tourism Office and stakeholders in analyzing tourist visit patterns and generating data-based tour package recommendations automatically and accessibly.

Early Warning System for the Impact of E-sports on Academics Based on Hybrid Naïve Bayes and Particle Swarm

2026-02-21T14:37:26+07:00

The integration of e-sports into the lifestyle of high school students in the era of industrial technology 4.0 brings a double dilemma between the development of digital skills and the risk of declining academic performance due to addiction. Previous research has successfully classified the impact of e-sports using the Naïve Bayes algorithm, but the model is static and only provides post-mortem analysis. This study proposes an Early Warning System (EWS) based on a hybrid of Naïve Bayes and Particle Swarm Optimization (PSO) designed to work dynamically and comprehensively. PSO is implemented to heuristically optimize attribute weights to overcome the weakness of the feature independence assumption in the pure Naïve Bayes algorithm. The test was carried out using the 10-fold cross-validation method on 178 student data of Madrasah Aliyah Negeri Rengasdengklok. The algorithm implementation resulted in an accuracy rate of 75.95% and an Area Under Curve (AUC) of 0.792. The main contribution of this research is the transformation from a traditional classification model to a proactive early warning system, where the system real-time monitors playing and studying duration, and then uses a probability threshold (Ƭ≥0.6) to trigger mitigation notifications to teachers and parents. In-depth analysis results show that behavioral variables such as playing duration have a much more massive level of significance (weight 1.0) compared to cognitive intelligence level or IQ (weight 0.001) in predicting academic failure. These findings provide a new paradigm for educational institutions in designing intervention strategies focused on time management and student digital literacy.

Implementation of the C4.5 Algorithm in the Website-Based Classification System for Students of Al-Ishlah Islamic Middle School

2026-02-21T22:29:57+07:00

Violations of school regulations constitute an issue that requires systematic management to support students’ character development; however, recording violations manually using books often creates difficulties in monitoring violation histories and determining appropriate sanctions. Therefore, this study aims to design and develop a web-based student violation classification system using the C4.5 algorithm. The classification process is carried out by constructing a decision tree based on entropy and information gain calculations to determine the best attribute. System performance evaluation is conducted using a confusion matrix along with accuracy, precision, recall, and F1-score metrics, with a focus on the test data. The results, based on 500 datasets with a 90:10 data split, indicate that the system is able to classify student violation levels very effectively, achieving an accuracy of 98.86% on training data and 97.96% on test data, with a precision of 0.91, recall of 1.00, and an F1-score of 0.95. With this strong performance, the system can serve as a supporting tool for recording violations and assisting decision-making in managing student violations more effectively and accurately.

Design of a Website-Based Climbing Equipment Registration and Rental Information System Using the Waterfall Method

2026-02-25T10:21:44+07:00

This research was motivated by the manual registration and equipment rental process at Lembah Lohe, Gowa Regency, which led to data recording errors, data loss risks, and low administrative efficiency. The objective of this study was to design a web-based hiking registration and equipment rental information system to improve data management effectiveness. The research applied the Waterfall method consisting of requirement analysis, system design, implementation, testing, and maintenance stages. The system was developed using PHP and MySQL and tested using Black Box Testing (Alpha and Beta). The results showed that all system functions operated according to specifications, and Beta testing produced an average index score of 82.40%, categorized as Very Good. Therefore, the developed system improves efficiency, accuracy, and accessibility of information for both hikers and administrators.

Development of an Absence System using IoT Technology Based on Radio Frequency Identification (RFID)

2026-02-26T20:12:36+07:00

SMK Negeri 1 Rambang Niru has implemented an online attendance system, but it still faces challenges related to reliability, accuracy, and efficiency, as the system is not based on Internet of Things (IoT) technology. This research aims to develop an automated attendance system utilizing IoT technology based on Radio Frequency Identification (RFID) to improve the effectiveness and efficiency of attendance tracking. The system uses RFID tags assigned to teachers and administrative staff , which are scanned by an RFID reader connected to a microcontroller. Once detected, the attendance data is immediately transmitted to a centralized server via an IoT communication module. The approach involves integrating RFID hardware with IoT infrastructure, as well as developing software for database management and user interface. The system was tested in the SMK Negeri 1 Rambang Niru work environment to evaluate scanning accuracy, response time, and data consistency. The results show that the RFID-based attendance system provides high reliability, fast data processing, and minimizes errors. This implementation demonstrates the potential to replace the existing online attendance system, enhancing data integrity and providing an efficient solution in modern workplace environments.

Comparison of TF-IDF and Word2Vec Feature Representations for Emotion Classification of Tokopedia E-Commerce Review Using LinearSVC

2026-02-25T22:17:12+07:00

This study aims to compare the performance of TF-IDF and Word2Vec feature representations for emotion classification of Tokopedia e-commerce reviews using the LinearSVC algorithm. The dataset used is PRDECT-ID, which consists of 5,400 Indonesian-language reviews labeled with positive and negative emotions. The preprocessing stages include case folding, non-alphabet character cleaning, slang normalization, stopword removal, Sastrawi stemming, and emoji handling. Feature extraction was performed using TF-IDF and Word2Vec, after which the models were trained using LinearSVC and evaluated through 5-Fold Cross Validation and holdout testing. The experimental results show that TF-IDF achieves better performance, with an accuracy of 0.65, a macro-F1 score of 0.645, and a Cohen’s Kappa value of 0.294. Meanwhile, Word2Vec attains an accuracy of 0.58 and a macro-F1 score of 0.540. These findings indicate that TF-IDF is more effective for short and informal texts characteristic of Indonesian e-commerce reviews.