Journal of Artificial Intelligence and Engineering Applications (JAIEA)

Analysis of User Acceptance of the Asempapak Village Information System Website Using the Technology Acceptance Model (TAM)

Thu, 15 Oct 2026 00:00:00 +0700

This study aims to analyze community acceptance of the Asempapak Village Information System website in Asempapak Village, Sidayu District, Gresik Regency, using the Technology Acceptance Model (TAM) approach. This research employed a quantitative method with primary data collected through questionnaires distributed to 92 residents of Asempapak Village. The research variables consisted of Perceived Usefulness, Perceived Ease of Use, Behavioral Intention to Use, and Actual System Use. The sampling technique used was simple random sampling, while the data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) with SmartPLS software. The results indicate that Perceived Ease of Use has a positive and significant effect on both Perceived Usefulness and Behavioral Intention to Use. In addition, Behavioral Intention to Use has a positive and significant effect on Actual System Use. However, Perceived Usefulness does not have a significant effect on Behavioral Intention to Use. These findings suggest that ease of use is the primary factor influencing the community's intention to use the village website. Therefore, website administrators should improve accessibility, enhance feature clarity, and provide continuous socialization and digital assistance to optimize the utilization of the village website.

Design and Implementation of a Single-Gate Parking Barrier Prototype using ESP32-CAM Based on the Internet of Things

Devandra Alifka Iswahyudi, Abdullah Hanif, Karyo Budi Utomo — Thu, 15 Oct 2026 00:00:00 +0700

The rapid proliferation of motorized vehicles within educational institutions necessitates robust parking management infrastructures to mitigate the unauthorized utilization of reserved spaces. This study proposes the design and implementation of an automated, Internet of Things (IoT)-based parking barrier prototype, employing an ESP32-CAM microcontroller as the core computing unit. The primary objective is to restrict facility access exclusively to authorized personnel by automating license plate identification and cross-referencing this telemetry with a cloud database. Unlike conventional edge-heavy systems, this architecture offloads Optical Character Recognition (OCR) processing to a cloud API to reduce thermal throttling and hardware latency on the ESP32-CAM. The methodology integrates an HY-SRF05 ultrasonic sensor for vehicle proximity detection, the ESP32-CAM for optical image acquisition, and a micro servo motor for barrier actuation. Upon detecting a vehicle within a 50 cm threshold, the system autonomously captures the license plate, processes the image via a cloud-based License Plate Recognition (LPR) API, and verifies the extracted string against a Firebase Realtime Database. Empirical testing demonstrated high system reliability; the ultrasonic sensor exhibited a spatial error margin of less than 2 cm, and the integrated IoT communication architecture achieved an operational latency of merely 2 to 3 seconds.

Analysis Effect of Perceived Ease of use and Perceived Usefulness on Halodoc Application User Satisfaction using the TAM Method

Nur Uswatun Hasanah, Vanesa Angelica Joceline, Riski Zulkarnain — Thu, 15 Oct 2026 00:00:00 +0700

The development of information technology has encouraged the transformation of healthcare services toward digital systems that are more effective and accessible. One of the most widely used digital healthcare platforms in Indonesia is Halodoc. This study aims to analyze the influence of Perceived Ease of Use (PEOU) and Perceived Usefulness (PU) on user satisfaction with the Halodoc application using the Technology Acceptance Model (TAM) approach. This study employed a descriptive qualitative method involving three active Halodoc users as informants. Data were collected through in-depth interviews, observation, and documentation. The findings indicate that perceived ease of use contributes positively to user satisfaction through a simple interface, intuitive navigation, and uncomplicated usage procedures. Perceived usefulness also influences user satisfaction, as the application helps users save time, facilitates access to healthcare services, and enables them to meet their healthcare needs more conveniently. Furthermore, both variables are interconnected in shaping user satisfaction. The easier and more useful the application is perceived to be, the higher the level of user satisfaction. These findings suggest that continuous improvements in usability and usefulness are necessary to maintain user satisfaction with digital healthcare applications.

Clustering Analysis of Civil Servant Data in Binjai City Based on Regional Work Unit and Age Using the K-Means Clustering Method

Sapriko, Novriyenni , I Gusti Prahmana — Thu, 15 Oct 2026 00:00:00 +0700

Civil Servants (Aparatur Sipil Negara/ASN) play a strategic role in government administration and public services in Binjai City. However, the management of ASN data has thus far remained largely administrative and descriptive, failing to adequately reveal employee distribution patterns based on regional work unit, age, and rank. This study aims to cluster ASN data in Binjai City using the K-Means Clustering algorithm based on the variables of regional work unit, age, and rank. The dataset consists of 498 civil servants obtained from the Personnel and Human Resource Development Agency (BKPSDM) of Binjai City, after removing anomalous data belonging to Rank I/d. The clustering process was carried out using MATLAB through the stages of data selection, transformation into numerical form, normalization, Euclidean Distance calculation, and iteration until the centroids stabilized. Testing was conducted on three cluster scenarios, namely k=3, k=4, and k=5, evaluated using Sum of Squared Error (SSE), the Davies-Bouldin Index, the Silhouette Score, and Variance analysis. The results show that k=4 produced the best clustering outcome, with a Silhouette Score of 0.5705, a Davies-Bouldin Index of 0.9309, an SSE of 29.9369, and a more balanced Variance distribution among clusters. The clustering results yielded four distinct groups of civil servants, one of which was identified as consisting predominantly of senior employees from the Education Office, requiring particular attention in succession planning. The findings of this study are expected to serve as a basis for policy-making in the management of human resources within the Binjai City Government.

Design and Development of an Ornamental Plant E-Commerce Mobile Application (Floramart) Using the Flutter and Laravel Frameworks

Thu, 15 Oct 2026 00:00:00 +0700

Currently, the sale of ornamental plants still heavily relies on conventional transactions and social media, resulting in unstructured order recording and limited market reach. This necessitates a digital transformation to improve operational efficiency and expand marketing areas. This study aims to design and develop a mobile e-commerce application prototype specifically for ornamental plants named Floramart. The development utilizes the Flutter framework on the frontend to produce a responsive cross-platform user interface. On the backend, the Laravel framework is used to securely manage product databases, transactions, and user authentication. The system development method used is the Prototype Method, which includes requirement gathering, system design, system building, and prototype evaluation. The result of this research is a Floramart application prototype integrated with a payment gateway (Midtrans) and automatic shipping cost calculation as well as tracking services Biteship. The implementation of this system is expected to provide transaction convenience for consumers and help ornamental plant MSMEs manage sales more effectively and systematically.

Design and Implementation of an IoT-Based Household Electricity Monitoring and Control System with Auto Cut-Off and Data Logger Features

Muhammad Ramadhan, Ahmad Rofiq Hakim, Fara Triadi — Thu, 15 Oct 2026 00:00:00 +0700

The growing demand for household electricity frequently leads to energy waste and overload risks. This study presents the design and implementation of an IoT-based Smart Energy system employing the ESP32 DevKit V1 microcontroller as the central controller. The system integrates an ACS758 current sensor, a ZMPT101B voltage sensor, an RTC DS3231 module, and a MicroSD card for timestamped historical data logging. Overload protection is achieved through a relay and a Miniature Circuit Breaker (MCB) that automatically disconnects the power supply when the load exceeds the configured threshold. The research adopts a Research and Development (R&D) methodology encompassing problem identification, system design, implementation, testing, and evaluation. Test results demonstrate that the system reliably monitors voltage, current, power, and energy consumption in real-time, records data to the SD card with accurate timestamps, and executes the auto cut-off function effectively. Comparative testing against a calibrated multimeter produce an average measurement error below 5.5%, confirming the system's suitability for household-scale energy monitoring applications.

Designing a Basic Digital Traceability Flow Model Sharia-Based in the Halal Industry Supply Chain (Case Study: Slaughterhouse and Beef Distribution Chain in the Pekalongan Region)

Thu, 15 Oct 2026 00:00:00 +0700

This study aims to design a basic digital traceability flow model aligned with Islamic Sharia principles within the halal industry supply chain, focusing on the animal slaughter sector at Slaughterhouses (RPH). The lack of transparent information regarding animal origin, slaughter processes, and meat distribution to consumers forms the background that drives the urgency of this research. The method used is mixed methods, combining a qualitative approach through direct observation at the RPH with a quantitative approach via questionnaires distributed to 30 respondents consisting of consumers and meat business operators. Field observation results indicate that the process at the RPH located in the Pekalongan region has met Sharia procedures in practice, but has not yet been equipped with a digital documentation system accessible to consumers. Questionnaire results show an average satisfaction score for halal information transparency needs of 4.18 out of 5, with 83.5% of respondents agreeing or strongly agreeing with the importance of supply chain information digitalization. The designed digital flow model covers four main stages: animal identification, Sharia verification, processing, and QR Code-based distribution. The system has not been fully implemented, but conceptual testing shows great potential in enhancing consumer trust in the halal status of products.

Keywords: digital traceability, halal supply chain, Sharia, slaughterhouse, information transparency, QR Code

Application of the Apriori Algorithm to Determine the Correlation of Tiktok Social Media use with the Learning Concentration of Senior High School Students

Feby Ria Syahpitri Br Ginting, Marto Sihombing, Tioria Pasaribu — Thu, 15 Oct 2026 00:00:00 +0700

Learning concentration is a crucial factor that supports the effectiveness of the learning process. Yet, the intensive use of short-video social media such as TikTok among students potentially disrupts their focus. At SMA/SMK Bina Bersaudara, students exhibit a high intensity of TikTok use, while the available student data has never been analyzed to reveal its relationship with learning concentration. This study aims to apply the Apriori algorithm, an association rule mining method in data mining, to identify patterns of relationship among student data, TikTok usage indicators, and learning concentration. The data were collected through questionnaires distributed to thirty students, transformed into binary form, and tested using RapidMiner with a minimum support of 20% and a minimum confidence of 80%. The analysis produced one valid association rule: students who open TikTok one to three times a day, most often at midday, and usually study for less than one hour per session tend to concentrate well, with a confidence level of 88.9%. The results indicate that the Apriori algorithm is able to reveal consistent relationship patterns between TikTok usage habits and learning concentration, which can serve as a data-based reference for schools, teachers, and parents in supervising social media use.

Camera-Based Parking Slot Monitoring System using Adaptive Brightness Thresholding and HSV Color Filtering

Muhammad Rifky Elviandy, Irwansyah, Agus Triyono — Thu, 15 Oct 2026 00:00:00 +0700

Traditional parking systems have been known to cause traffic queues and difficulty in locating parking spaces. This study proposes a camera-based parking slot monitoring system that combines Region of Interest (ROI), adaptive brightness thresholding, and HSV color detection, without relying on any artificial intelligence algorithm such as CNN or YOLO. This approach was chosen so that the system can run lightly in real time on a low-power embedded device such as a Raspberry Pi 4, without requiring a training dataset or GPU acceleration. A USB camera was used as the visual sensor, a Raspberry Pi 4 as the computing unit, Firebase Realtime Database as the data storage, and a web-based dashboard as the monitoring interface. A small-scale parking prototype consisting of 20 slots was tested under two lighting conditions. As a result, detection accuracy reached 87.5% under bright lighting conditions and 65% under low-light conditions, with an average data update latency of 1–2 seconds. These findings indicate that a rule-based, pixel-brightness approach is sufficiently effective for small-scale parking monitoring, although performance under low-light conditions still requires improvement through additional lighting or more adaptive thresholding methods.

Web-Based Geographic Information System (GIS) for Village Asset Management in Sungai Raya Village

Muhhamad Zulkifli, Dwi Yuli Prasetyo, Bayu Rianto — Thu, 15 Oct 2026 00:00:00 +0700

The development of information technology has led to the emergence of various systems to assist in managing spatial data, one of which is Geographic Information System (GIS). GIS offers a digital and efficient solution for asset management, especially in areas that lack integrated data systems, such as Sungai Raya Village, Batang Tuaka Subdistrict, Indragiri Hilir Regency. Currently, asset data collection in the village is not conducted systematically, which hinders the identification, monitoring, and data-based decision-making processes. This study aims to design and develop a web-based geographic information system that can digitally map village assets and visually display the information using map imagery. The system is developed using PHP programming language and MySQL database, following the waterfall model in the system development life cycle. The resulting system is expected to improve efficiency in asset identification, simplify maintenance processes, and support accurate and transparent decision-making. By implementing GIS for asset management, Sungai Raya Village is expected to manage its resources more optimally and sustainably.

Risk Analysis of Halal Compliance in the Halal Industry Supply Chain

Raihan Akmal Taufiq, Ghoniu Rozaq; Zimam Asshidqi; Fatkhur Rokhman — Thu, 15 Oct 2026 00:00:00 +0700

The growth of the halal industry requires every business operator to ensure the halal status of their products at every stage of the supply chain. This study aims to analyze halal risks in the supply chain of fried chicken MSMEs using a simple risk mapping approach. The study employs a mixed-methods approach based on library research, combining qualitative analysis to identify potential halal risks with simple quantitative analysis through the assessment of likelihood, impact, and risk scores. The study focuses on three main stages of the supply chain: raw material procurement, the production process, and product distribution. The results indicate that the highest-level risks stem from the use of chicken not slaughtered in accordance with Islamic law, the use of food additives with unclear halal status, suppliers lacking halal certification, and the occurrence of cross-contamination during the production process. Based on the risk mapping results, these risks fall into the high category and therefore require control as a top priority. In contrast, risks in the distribution stage tend to fall into the moderate to low categories. The application of risk mapping provides an overview of critical halal control points while helping to systematically determine risk control priorities. The findings of this study are expected to serve as a reference for MSMEs in improving halal risk management and supporting the implementation of a more effective and sustainable halal supply chain.

Design and Construction of a Multifunctional Educational Mobile Robot for STEM Learning Based on Visual Programming Using ESP32

Mochamad Farid Fadhlulloh, I Gede Susrama Mas Diyasa, Basuki Rahmat — Thu, 15 Oct 2026 00:00:00 +0700

The limited availability of STEM learning media capable of interactively and affordably integrating cross-disciplinary concepts poses a significant barrier to STEM education implementation in Indonesian secondary schools. This study proposes and realizes a Multifunctional Educational Mobile Robot based on the ESP32 microcontroller that can be visually programmed using PictoBlox—a Scratch-based platform—as a solution to this gap. The system is designed to operate in four modes: Line Follower (TCRT5000 line sensor), Obstacle Avoidance (HC-SR04 ultrasonic sensor), analog joystick control, and Android smartphone control via Bluetooth using the Dabble application. An aluminum chassis with Mecanum wheels enables omnidirectional movement that enriches the coverage of vector physics concepts. Systematic testing yielded quantitative data: the HC-SR04 sensor achieved a stable error of 2.80% across the 10–80 cm; the TCRT5000 sensor achieved 96.7% accuracy under standard conditions; Bluetooth communication maintained a latency of 118–122 ms with zero packet loss at 1–8 meters; and the success rates for Line Follower, Obstacle Avoidance, and Smartphone Control modes were 93.3%, 96.0%, and 97.0%, respectively. The system demonstrated full reliability during continuous 30-minute sessions. Pedagogical analysis confirmed that the four modes collectively cover all STEM domains, making this robot a comprehensive STEM learning media candidate for senior and vocational high school levels.

Plant Growth Monitoring System for Scallion

Hangga Aditya Nugraha, Abdullah Hanif, Wahyuni Eka Sari — Thu, 15 Oct 2026 00:00:00 +0700

This study aims to address inconsistencies in urban farming plant care by designing a standalone automation system for green onions using the ESP32 microcontroller. The system operates without an internet connection, implementing a hybrid watering logic adaptive to plant age: once daily during the initial phase (days 1-10) and automatically increasing to twice daily in the advanced phase (days >10), alongside an emergency feature triggered when soil is detected as extremely dry (ADC > 2600). Additionally, a smart lighting system using growlights activates automatically during dark daytime conditions. Test results demonstrate that the system executes all scenarios accurately and stably, providing an efficient care solution with minimal human intervention.

A Comparative Analysis of User Experience Between the DANA and GoPay Mobile Applications Using the User Experience Questionnaire (UEQ)

Karina Maretasari, Pradita Imaniar, Riski Zulkarnain, Nasruddin Bin Idris — Thu, 15 Oct 2026 00:00:00 +0700

This study aims to analyze and compare the quality of the user experience provided by the two most popular digital wallet applications in Indonesia, DANA and GoPay. A quantitative comparative approach was employed by distributing the User Experience Questionnaire (UEQ) to active respondents who regularly use both platforms. User experience was systematically evaluated using the six UEQ dimensions to objectively assess both pragmatic and hedonic quality. The findings indicate that both applications achieved positive evaluations, with mean scores exceeding the recommended UEQ benchmark across all dimensions. GoPay demonstrated superior performance in the pragmatic quality dimensions, particularly Efficiency and Dependability, reflecting its stable transaction infrastructure and efficient payment processes. In contrast, DANA outperformed GoPay in the hedonic quality dimensions, especially Stimulation and Novelty, owing to its dynamic user interface and the successful integration of interactive gamification features. The study concludes that pragmatic quality plays a more influential role in shaping users' overall impressions of GoPay, whereas DANA's strength lies in delivering a more engaging and innovative user experience that supports long-term user retention. These findings provide practical insights for digital wallet developers seeking to optimize interaction design by balancing pragmatic and hedonic aspects of user experience

Implementation of Web-Based POS and Operational Laundry System at Laundry Melati

Muhammad Ainal Fikri Lubis, Eva Rahmawati, Sri Diantika — Thu, 15 Oct 2026 00:00:00 +0700

The operational management at Laundry Melati currently faces several challenges, including manual transaction recording, vulnerability to human errors, delayed customer notifications, and unintegrated digital payment verification. This research aims to design and develop a web-based laundry operational and Point of Sale (POS) information system to optimize these business processes. The system was developed leveraging the Laravel 12 framework and MySQL for database management. Furthermore, external API gateways were integrated, including Midtrans for automated digital payment processing and WhatsApp gateway for real-time customer notifications. System architecture was modeled through Unified Modeling Language (UML) diagrams, including Use Case, Activity, and Class diagrams, alongside Entity Relationship Diagrams (ERD) and Logical Record Structures (LRS) to ensure a robust multi-role access control for Admin and Cashier roles. The system underwent evaluation using Black-box testing methods. The results demonstrate that all functional components, including automated transaction billing, real-time status updates, and digital payment callbacks, operate successfully with zero critical defects. In conclusion, the implemented system effectively enhances data validity, minimizes transaction processing time, and provides precise automated financial reporting.

Design and Construction of a Website-Based Information System on the Potential of Cermo Village using the Rad Method

Arthur Ashley Christy Ong — Thu, 15 Oct 2026 00:00:00 +0700

Information Systems are a collection of interconnected components that work together to collect, process, store, and distribute data. In this context, Cermo Village has a tourist spot called Wisata Taruloyo. In addition, Cermo Village also produces several bamboo woven products in various forms. In this case, there is a problem in the promotion sector, especially in the tourism sector of Cermo Village. The lack of public knowledge about Wisata Taruloyo in Cermo Village results in fewer visitors than expected, and the results are less than optimal. To address this problem, an information system for Cermo Village's potential was created and implemented in the form of a website, which can be accessed online using a browser. The development method used is Rapid Application Development (RAD). With the Cermo Village potential information system, it is hoped that the village's potential can be maximized.

Sales Prediction at Caffe Resto Omah Lawas Using the Naive Bayes Method

Putra Vira Madya, Munif — Thu, 15 Oct 2026 00:00:00 +0700

This study discusses the design and implementation of a sales prediction system for Caffe Resto Omah Lawas using the Naïve Bayes method. The objective of this research is to assist business owners in anticipating fluctuations in customer demand, optimizing product inventory, and improving data-driven marketing strategies. The research process includes sales data collection, data preprocessing to clean and prepare the dataset, transformation of numerical attributes into categorical attributes, and dataset splitting into 80% training data and 20% testing data. The system was developed using the PHP programming language and MySQL database and implemented as a web-based application to facilitate accessibility and ease of use. The experimental results show that the proposed system achieved an accuracy of 90%, with excellent performance in the "Best-selling" and "Unsold" classes, while the performance for the "Normal" class was relatively lower. Furthermore, the system provides strategic recommendations based on prediction results, including inventory adjustment, promotional planning, and product development according to customer preferences. Therefore, the implementation of the Naïve Bayes method has proven to be effective as a data-driven decision support tool for sales and marketing management.

Analysis of the Acceptance of the Indonesia Pintar Programme Website using the Technology Acceptance Model (TAM) Among Administrative Staff at Pangeran Antasari Vocational High School Balikpapan

Thu, 15 Oct 2026 00:00:00 +0700

Advances in information technology have driven the digitisation of educational services through the Indonesia Pintar (PIP) Programme website to support the administration of educational assistance in schools. However, its implementation still faces challenges, such as an unreliable login system, a lack of awareness-raising, and differences in user perceptions. This study aims to analyse the level of acceptance of the Indonesia Pintar (PIP) Programme website amongst administrative staff at SMK Pangeran Antasari in Balikpapan using the Technology Acceptance Model (TAM) approach. The study employed a descriptive qualitative method utilising interviews, observation and documentation, which were subsequently analysed using the Miles and Huberman model. The results indicate that the website has been accepted by users and assists with the administration and management of beneficiary data. In terms of Perceived Ease of Use, the website is considered reasonably easy to use, although there is still a need for improved awareness-raising and feature enhancements. Regarding Perceived Usefulness, the website is seen to improve work efficiency, although some administrative staff still use Microsoft Excel. Consequently, improvements to the systems’s quality and user training are required to optimise the website’s implementation.

Design and Development of an IoT-Based Student Attendance Monitoring System Using RFID and ESP32-CAM Integration with Telegram Notifications for Parents

Rizky Kurniawan, Husnul Khair, Muammar Khadapi — Thu, 15 Oct 2026 00:00:00 +0700

Abstract

The development of Internet of Things (IoT) technology has created opportunities to improve the effectiveness of attendance systems in schools. Conventional manual attendance processes have several limitations, including inefficiency, susceptibility to recording errors, and delays in delivering attendance information to parents. This study aims to design and develop an IoT-based student attendance monitoring system by integrating RFID and ESP32-CAM with Telegram notifications for parents. The system utilizes RFID for student identification, ESP32-CAM to capture attendance documentation, and a Telegram Bot to automatically send attendance notifications to parents. The system was developed using the Prototype development method, which consists of requirements analysis, system design, prototype development, testing, and evaluation until the system meets user requirements. The system is also supported by a web-based application used to manage student, teacher, parent, class, and attendance report data. The results show that the proposed system is capable of automatically recording student attendance, displaying attendance data in real time, capturing images during the attendance process, and sending attendance notifications to parents via Telegram. Therefore, the developed system improves the efficiency of the attendance process, facilitates real-time student attendance monitoring, and strengthens communication between schools and parents.

Keywords: Internet of Things (IoT), RFID, ESP32-CAM. Attendance System, Telegram Bot

Design and Implementation of an Internet of Things (IoT)-Based Smart Farming System for Chili Plants in Parit 4, Tembilahan Hulu

Eki Pirnanda, Bayu Rianto, Muh Rasyid Ridha — Thu, 15 Oct 2026 00:00:00 +0700

Efficient irrigation management is essential for improving chili cultivation productivity while reducing excessive water consumption. This study aimed to design and implement an Internet of Things (IoT)-based smart farming system capable of performing automatic irrigation and real-time environmental monitoring. The proposed system employed an ESP32 microcontroller integrated with a capacitive soil moisture sensor, a DHT22 sensor, a relay module, and a DC water pump. Environmental data were transmitted through a Wi-Fi network to a MySQL database and visualized using a web-based dashboard, allowing users to monitor soil moisture, air temperature, and relative humidity while controlling irrigation remotely. The developed system was evaluated through Black Box Testing to verify hardware and software functionality, communication reliability, and automatic irrigation performance. The implementation results demonstrated that all system components operated successfully, enabling stable real-time monitoring, automatic irrigation based on predefined soil moisture thresholds, and remote manual control through the web interface. The communication process between the ESP32 and the web server achieved response times below three seconds, while the usability evaluation obtained a feasibility score of 88.8%, indicating that the system is highly suitable for practical application. The proposed IoT-based smart farming system effectively improves irrigation efficiency, optimizes water utilization, and supports sustainable chili cultivation through integrated environmental monitoring and automated irrigation management.

Selection of Toddlers at Risk of Stunting in Binjai City using the MOORA Method

Kelvinsyah Reynaldi, Relita Buaton, Suci Ramadani — Thu, 15 Oct 2026 00:00:00 +0700

Stunting is a growth and development disorder in toddlers caused by long-term malnutrition, which can occur from pregnancy to 24 months of age. The Binjai City Population Control and Family Planning Office has an important role in efforts to prevent and handle stunting, but the selection of toddlers at risk of stunting still faces obstacles because the assessment process is still carried out manually, so it becomes less effective and takes a long time so that it has the potential to cause errors in decision-making. Therefore, this study aims to build a Decision Support System (SPK) that is able to help the Binjai City PPKB Office determine the selection of toddlers at risk of stunting objectively, quickly, and accurately. The method used in this study is MOORA (Multi-Objective Optimization on the Basis of Ratio Analysis) which is a method that has the ability to solve multicriteria decision-making problems through the process of normalization, weighting, and calculation so as to produce alternative rankings based on priority levels. The research data was obtained through observations, interviews, and documentation at the Binjai City Population Control and Family Planning Office. The results of the study show that the MOORA method is able to produce a ranking of toddlers based on the level of stunting risk in a systematic and objective manner, so that it can help the Binjai City PPKB office for decision-makers and increase the effectiveness of prioritizing toddlers who need further handling. The Implementation of the Selection of Toddlers at Risk of Stunting The MOORA Decision Support System Method is expected to support stunting prevention and control programs in a more targeted manner.

Analysis of the Relationship Pattern Between Extracurricular Participation and Student Attendance Towards Academic Achievement using the Apriori Algorithm (Case Study: SMAS Harapan Hamparan Perak)

Najwa Salsabila Damanik, Marto Sihombing, Kristina Annatasia Br Sitepu — Thu, 15 Oct 2026 00:00:00 +0700

Student development evaluation at Harapan Hamparan Perak Senior High School has been conducted manually without systematic pattern analysis. This study analyzes the relationship patterns between extracurricular participation, attendance levels, and academic achievement using the Apriori algorithm. Data from 545 students in grades X, XI, and XII were processed using WEKA 3.8.6 with a minimum support of 0.2 and minimum confidence of 0.6. The results produced two main association rules: High Achievement → High Attendance (confidence 74%) and High Attendance → High Achievement (confidence 70%), proving a strong reciprocal relationship between attendance discipline and academic achievement. Meanwhile, extracurricular variables did not appear as a dominant pattern

Development of an IoT-Based Smart Medication Dispenser with Real-Time Monitoring to Improve Patient Medication Adherence

Rafi Abdurrohim, Abdullah Hanif, Karyo Budi Utomo — Thu, 15 Oct 2026 00:00:00 +0700

Patient medication adherence in Indonesia remains a serious challenge, with reported non-adherence rates reaching 57% among chronic-disease patients, while conventional manual supervision by medical staff is time-consuming and prone to human error. This study aims to design and build a prototype of an Internet of Things (IoT)-based Smart Medication Dispenser equipped with real-time monitoring to improve patient medication adherence in a hospital environment. The system is built around an ESP32 microcontroller integrated with a DS3231 real-time clock for schedule accuracy, an RFID module for staff and patient authentication, a servo motor as the dispensing actuator, an ultrasonic sensor for retrieval verification, a DHT22 sensor for storage-condition monitoring, and a buzzer with an LCD for audio-visual notification. Activity data is transmitted to Google Firebase and displayed on a web dashboard for remote monitoring. The system was evaluated using the black-box testing method across seven functional scenarios covering initialization, staff configuration access, schedule notification, patient verification, retrieval confirmation, and timeout handling. The results show that all tested scenarios executed successfully: the system accurately dispensed medication according to schedule, restricted configuration access to authenticated staff only, and reliably logged and transmitted retrieval status—including "Successfully Taken", "Skipped", and "Not Taken" conditions—to the dashboard in real time. These findings indicate that the developed prototype can serve as a practical and reliable technological solution to support medication-adherence monitoring in clinical settings.

Design and Development of a Mobile-Based E-Commerce Application Using Flutter Framework for QOF Drink MSME with the Waterfall Method

Thu, 15 Oct 2026 00:00:00 +0700

The increasing adoption of digital technology has encouraged micro, small, and medium enterprises (MSMEs) to utilize mobile-based information systems to improve operational performance, broaden market opportunities, and enhance customer service quality. Despite these developments, QOF Drink continues to manage product promotion, order processing, transaction recording, and delivery coordination through conventional methods, leading to operational inefficiencies and limited service accessibility. Therefore, this study aims to design and develop a mobile-based e-commerce application for QOF Drink by utilizing the Flutter framework and the Waterfall software development methodology. The development process includes requirements analysis, system design, implementation, system testing, and maintenance planning. Flutter was employed to build a responsive mobile application, while Firebase was integrated as the backend platform to handle user authentication, real-time database services, and transaction synchronization. The developed application combines customer, administrator, and driver functionalities into a single platform, providing features such as product catalog management, shopping carts, checkout, order history, delivery assignment, and real-time order tracking. The implementation results indicate that the proposed application streamlines business operations, improves transaction management, enhances customer convenience, and supports the digital transformation of QOF Drink through an integrated mobile-based e-commerce platform.

Analysis of Downtime Causes and Proposed Belt Conveyor Performance Improvement Using Root Cause Analysis and Plan-Do-Check-Action Methods at PT X

Muhammad Rafi Ardyanto, Sinta Dewi — Thu, 15 Oct 2026 00:00:00 +0700

This study analyzes the causes of downtime and proposes performance improvement solutions for Belt Conveyor machines at PT X using an integrated approach of Root Cause Analysis (RCA) and Plan-Do-Check-Action (PDCA) methods. Operating in the tobacco processing industry in Mojokerto, East Java, PT X faces significant production inefficiency challenges, with Belt Conveyors (M-08) contributing 47.09% of total downtime (103.93 hours) in March 2025. The research employs Pareto Diagram to identify priority problems and Fishbone Diagram to map root causes across six factors: Machine, Method, Man, Material, Measurement, and Environment. Analysis reveals that the dominant causes are jammed roller idlers and motor overheating due to tobacco dust accumulation. The 5W+1H method is utilized to formulate corrective and preventive actions, including upgrading bearings to Rubber Sealed type (6204-2RS) and establishing systematic motor cleaning procedures. Implementation results project a 50% downtime reduction from 103.93 hours to 52 hours, with machine availability increasing from 98.51% to 99.26%. The study demonstrates that integration of RCA and PDCA provides an effective framework for identifying root causes and implementing sustainable maintenance solutions. Standardization of component specifications and cleaning procedures is established to ensure continuous improvement and optimal operational performance in tobacco processing operations.

Classification of East Sumba Weaving Motifs Based on Texture Using GLCM and SVM Methods

Rambu Rinta Okrin Malo — Thu, 15 Oct 2026 00:00:00 +0700

East Sumba woven fabric represents a cultural heritage of significant aesthetic and philosophical value, expressed through its diverse array of motifs. However, manual identification of these motifs is prone to high subjectivity, inconsistency, and time-consuming processes. This study aims to develop an automated classification system for East Sumba woven fabric motifs based on texture characteristics, utilizing a combination of the Gray Level Co-occurrence Matrix (GLCM) method and the Support Vector Machine (SVM) algorithm. The dataset comprises 500 digital images of woven fabrics covering five motif variations Baby Crocodile, Chicken, Bird, Dragonfly, and Horse with 100 samples for each. The research workflow encompasses image preprocessing (grayscale conversion), GLCM texture feature extraction (using contrast, correlation, energy, and homogeneity parameters), and classification modeling via an RBF kernel-based SVM algorithm. The dataset was split into 400 training images and 100 testing images. System performance was evaluated using a confusion matrix. The model achieved excellent overall performance, yielding an accuracy of 85.00%, macro-precision of 91.43%, macro-recall of 85.00%, and a macro-F1 score of 82.55%. A slight drop in accuracy was attributed to classification ambiguity in the Chicken motif class, which shares highly similar texture feature values with the Bird motif. Nevertheless, the study demonstrates that the integration of GLCM and SVM methods is highly effective and suitable for implementation to support the automation, digitization, and preservation of East Sumba's cultural heritage.

Visual Analysis of Facial Features using a Convolutional Neural Network (CNN) to Provide Styling Recommendations

Thu, 15 Oct 2026 00:00:00 +0700

This study aims to develop a facial shape analysis system that provides hairstyle recommendations based on face type using the Convolutional Neural Network (CNN) method. The dataset consists of face images categorized into four labels: heart, oblong, round, and square. The preprocessing process includes image resizing and normalization to improve model accuracy. The MobileNetV2 architecture is employed for feature extraction and classification due to its efficiency in handling image-based recognition tasks. The model achieved an accuracy of 77%, indicating that it can effectively classify face shapes and generate hairstyle recommendations. The system is deployed as a web application, allowing users to upload their photos, automatically identify face shapes, and receive hairstyle suggestions that best match their facial characteristics. This research demonstrates the potential of CNN in implementing intelligent styling recommendations based on visual facial features.

Fake News Classification Using the Multinomial Naive Bayes Algorithm with TF-IDF Weighting on the FakeNewsNet Dataset

Deny Pratama Silaban, Pesta Purba, Jaya Tata Hardinata — Thu, 15 Oct 2026 00:00:00 +0700

The proliferation of digital and social media platforms has accelerated the production and dissemination of fake news, creating serious risks to public trust, democratic discourse, and social stability. Automatic detection of fake news at the point of publication is therefore an urgent research problem in text mining and natural language processing. This study develops and evaluates a fake news classification pipeline based on the Multinomial Naive Bayes (MNB) algorithm combined with Term Frequency-Inverse Document Frequency (TF-IDF) feature weighting, using the FakeNewsNet benchmark corpus of news article titles labeled as real or fake. After parsing and cleaning the raw repository, 21,637 unique, non-empty news titles remained, comprising 16,399 real (75.8%) and 5,238 fake (24.2%) instances. The text was preprocessed through case folding, URL removal, numeral removal, punctuation removal, tokenization, stopword removal, and lemmatization prior to TF-IDF vectorization with a 5,000-term vocabulary. The dataset was partitioned using an 80:20 stratified train-test split (17,309 training and 4,328 testing instances), and a Multinomial Naive Bayes classifier was trained on the resulting TF-IDF matrix. On the held-out test set, the model achieved an overall accuracy of 82.62%, a weighted precision of 81.98%, a weighted recall of 82.62%, and a weighted F1-score of 80.41%. Class-level analysis revealed a pronounced asymmetry: the real-news class was recognized with high precision (83.28%) and recall (96.43%), whereas the fake-news class, being the minority class, achieved a lower recall of 39.41% despite a precision of 77.92%. These findings confirm that TF-IDF-weighted Multinomial Naive Bayes is a computationally efficient and reasonably accurate baseline for short-text fake news classification, while also highlighting the class-imbalance challenge that limits fake-news recall. The paper discusses the algorithmic rationale, the practical trade-offs of the approach, and directions for improving minority-class detection in future work.

Analysis of Digital Technology Acceptance in Supporting Community Digital Literacy Using the Technology Acceptance Model in Balikpapan City

Thu, 15 Oct 2026 00:00:00 +0700

Digital technology has become an essential part of modern society, providing easier access to information, communication, education, and public services. Despite the rapid growth in digital technology adoption, differences in individuals' digital literacy remain a significant challenge, particularly in their ability to search, evaluate, and utilize digital information effectively. Understanding the factors influencing digital literacy is therefore essential for developing effective digital transformation strategies. This study aims to analyse the influence of digital technology acceptance on community digital literacy in Balikpapan City by applying the Technology Acceptance Model (TAM). Specifically, the study examines the effects of Perceived Usefulness (PU) and Perceived Ease of Use (PEOU) on Digital Literacy (DL). A quantitative research design was employed using an online questionnaire distributed through Google Forms. A total of 105 valid responses were collected from residents of Balikpapan City. The collected data were analysed using descriptive statistics, validity and reliability tests, and multiple linear regression analysis. The findings reveal that both Perceived Usefulness and Perceived Ease of Use have positive and statistically significant effects on Digital Literacy. The regression analysis indicates that the proposed model explains 76.0% of the variance in digital literacy (R² = 0.760). Furthermore, Perceived Ease of Use demonstrates a stronger influence than Perceived Usefulness, suggesting that technologies that are easy to understand and operate encourage greater engagement with digital platforms and contribute more effectively to improving digital literacy. This study provides empirical evidence supporting the applicability of the Technology Acceptance Model in explaining community digital literacy. The findings offer practical implications for local governments, educational institutions, and policymakers in designing digital literacy programs by emphasizing not only the usefulness of digital technologies but also their accessibility and ease of use.

Implementation of the SMARTER Method for Ranking Student Discipline Levels in a Web-Based Student Violation Management Information System

Muhamad Adnan Alwi, Resti Amalia — Thu, 15 Oct 2026 00:00:00 +0700

Discipline management for students (santri) in Islamic boarding schools is commonly still handled manually, which creates risks of lost records, difficulty tracing violation history, and slow report recapitulation. This study designs and implements a web-based information system for managing student violations at Pondok Pesantren Bait Qur'ani Bogor, integrating the Simple Multi-Attribute Rating Technique Exploiting Ranks (SMARTER) method to rank student discipline levels objectively. Criterion weights were derived using the Rank Order Centroid (ROC) formula from four criteria — point accumulation, violation level, violation type, and violation category — producing weights of 52.08%, 27.08%, 14.58%, and 6.25% respectively. The system was developed following the Waterfall model and built with the Laravel framework and a MySQL database. Black-box testing across ten use cases and 28 scenarios achieved a 100% success rate, while a Likert-scale user acceptance test involving eight respondents yielded an average acceptance score of 79.73%, categorized as “Good.” Manual SMARTER calculations for five sample students matched the system's automated ranking exactly, confirming the correctness of the algorithm implementation. The resulting system successfully replaces manual, paper-based recording with a real-time, objective ranking mechanism that supports pesantren supervisors in identifying students who require special guidance.

Web-Based Cashier Information System with Royalty Feature at FA Playground

Habib Al Huda, Bayu Rianto, Samsudin — Thu, 15 Oct 2026 00:00:00 +0700

FA Playground is a children's play-area business located in Tembilahan City. Sales transaction records and bookkeeping at this business are still carried out conventionally, which frequently causes customer-data duplication, damaged or lost vouchers, and difficulty for cashiers in tracing customer transaction history. This research aims to design and build a web-based cashier information system equipped with a royalty feature, which grants customers a bonus play session after reaching a certain number of visits, in order to increase customer retention. The system development method used is the Waterfall model, with the requirement-analysis stage using the PIECES (Performance, Information, Economics, Control, Efficiency, Service) approach, system design using the Unified Modeling Language (UML), implementation using the PHP programming language and MySQL database, and testing using the black-box method covering functionality and usability aspects. The functionality test on 37 system functions involving 10 respondents showed a success rate of 100% (X=1), while the usability test obtained a feasibility percentage of 96.86%, which falls into the very good category. Thus, this web-based cashier information system with a royalty feature has proven capable of speeding up the transaction process, reducing customer-data duplication, securing voucher data, and simplifying bookkeeping and sales reporting at FA Playground.

Sentiment Analysis of Students Toward Academic Services at Universitas Kristen Wira Wacana Sumba Using the SERVQUAL Method

Kristanti Rambu Banja Uru, Pingky Alfa Ray Leo Lede, Raynesta Mikaela Indri Malo — Thu, 15 Oct 2026 00:00:00 +0700

Academic service quality is an important factor influencing student satisfaction in higher education institutions. Regular evaluation is needed to assess whether the services provided meet student expectations. This study aims to analyze the quality of academic support services at Universitas Kristen Wira Wacana Sumba using the Service Quality (SERVQUAL) method combined with sentiment analysis. The SERVQUAL method measures the gap between student perceptions and expectations based on five dimensions of service quality: Tangibles, Reliability, Responsiveness, Assurance, and Empathy. Sentiment analysis is applied to students’ open-ended comments to identify positive, neutral, and negative opinions. This study employed a descriptive quantitative approach with purposive sampling involving 30 active students. The results show that all SERVQUAL dimen Academic sions have negative gap values with an average of -0.37, indicating that the services have not fully met students' expectations. The responsiveness dimension shows the largest gap (-0.57), followed by reliability (-0.54), making them the main priorities for improvement. Sentiment analysis results indicate that 70% of students expressed positive opinions, 20% expressed neutral opinions, and 10% expressed negative opinions. The integration of SERVQUAL and sentiment analysis provides a comprehensive evaluation by combining quantitative measurements of service quality with qualitative insights into students’ perceptions.

Development of a GPS-Based Digital Attendance System with Attendance Validation in a Learning Management System using the Rapid Application Development Method (Case Study: Digitech University)

Liani Siti Lutfiah, Fadzar Rusghana, Ridha Adjie Eryadi — Thu, 15 Oct 2026 00:00:00 +0700

Digital attendance systems that still rely on manual confirmation have the potential to cause inaccuracies in attendance data and open the door to fraud. This study aims to develop a GPS geofencing-based digital attendance system using the Rapid Application Development (RAD) method within the DIgitech University Learning Management System (LMS). The research method used is descriptive, with a system development approach following the RAD stages. The system was developed as a web-based prototype with a multi-factor validation mechanism to verify student attendance based on location and time. Test results show that the digital attendance system improves the accuracy of attendance records, reduces the potential for fraud, and enhances the efficiency of the attendance process for both faculty and students.

Internet of Things-Based Tobacco Dryer Monitoring System Design

Risma Tamsar, Karyo Budi Utomo, Agus Triyono — Thu, 15 Oct 2026 00:00:00 +0700

This research aims to design and implement an Internet of Things (IoT)-based tobacco dryer monitoring system capable of automatically regulating temperature and humidity while displaying data in real-time through a web dashboard. The system is built using an ESP32 microcontroller as the main controller, a DHT22 sensor for temperature and humidity readings, and a Load Cell sensor for tobacco weight measurement. The actuators consist of a heating lamp and a 12V DC fan controlled via a 2-channel relay module. Test results show that the DHT22 sensor achieves an average temperature reading deviation of ±0.25°C and humidity of ±1%, while the Load Cell records an average error of 0.6%. The automatic control system maintains the temperature within the range of 35°C–45°C with a response time of 1–2 seconds. Temperature, humidity, and weight data are successfully transmitted and displayed on the web dashboard in real-time with a data refresh interval of 1–2 seconds. The system is proven to be stable, accurate, and efficient in supporting the tobacco drying process independently of weather conditions.

IoT-Based Soil Moisture Monitoring and Drip Irrigation System for Chili Plants Using NodeMCU and Blynk

Desy Fatmalia Putri, Arnes Sembiring, Ratih Puspadini, Saifuddin Muhammad Jalil — Thu, 15 Oct 2026 00:00:00 +0700

Water management is one of the most important factors influencing the growth and productivity of chili plants. Conventional irrigation methods often result in inefficient water consumption because watering is performed manually without considering actual soil moisture conditions. This study aims to develop an Internet of Things (IoT)-based drip irrigation system capable of monitoring soil moisture in real time using NodeMCU ESP8266 and the Blynk application. The proposed system integrates four YL-69 soil moisture sensors, a DHT22 temperature sensor, an I2C LCD, a relay module, and a DC water pump. Median Filtering is implemented to reduce fluctuations in soil moisture measurements before controlling the irrigation process. The system automatically activates the water pump when soil moisture is below 60% and stops irrigation when soil moisture exceeds 80%. Experimental results demonstrate that the developed system successfully performs automatic irrigation, provides real-time monitoring through both the I2C LCD and Blynk platform, and improves the stability of sensor readings using Median Filtering. Therefore, the proposed system can support efficient water management and smart agriculture applications for chili cultivation.

Design and Development of an Internet of Things (IoT)-Based Water Level Monitoring System for Flood Early Warning at Lempake Dam

Soni Heri Saputra, Ahmad Rofiq Hakim, Fara triadi — Thu, 15 Oct 2026 00:00:00 +0700

Floods are among the most frequent natural disasters in Indonesia, often causing significant damage to communities and infrastructure. Monitoring the water level at the Lempake Dam is still carried out manually, resulting in delays in obtaining early flood information. This study aims to design an Internet of Things (IoT)-based water level monitoring system as an early flood warning tool. The system uses an ESP32 microcontroller and a JSN-SR04T ultrasonic sensor to measure water surface levels in real time. Measurement data are displayed on a 16x2 I2C LCD and sent to Firebase for visualization on a web dashboard. The system is also equipped with LED indicators, a buzzer, and an automatic notification feature via Telegram when the water level exceeds the safety limit. The test results show that the system operates stably with a measurement accuracy of ±3 mm and provides rapid warning notifications < 2 second. Therefore, this system can assist operators in efficiently monitoring water conditions and support flood disaster mitigation efforts at the Lempake Dam.

Clustering Students' Level of Understanding of Information and Communication Technology Subjects using the K-Means Algorithm

Adinda Febrina, Suci Ramadani, Imeldawaty Gultom — Thu, 15 Oct 2026 00:00:00 +0700

Information and Communication Technology (ICT) within the Merdeka Curriculum requires students not only to operate technology but also to understand basic computational concepts and problem-solving, so its success depends heavily on students' level of understanding. The problem is that at SMA Negeri 3 Binjai, students' understanding is uneven, and the available academic score data has not been processed to map patterns of understanding, making it difficult for teachers to determine appropriate learning interventions. As a solution, this study applies the K-Means Clustering algorithm to 634 student score records covering practicum assignment scores, midterm (UTS) scores, and final-exam (UAS) scores, which are transformed into categorical codes based on the Learning Objective Achievement Criteria (KKTP), standardized, and then grouped through the CRISP-DM stages using the Python programming language. The number of clusters is tested over the range K = 2 to 10 using the Silhouette Score and the Davies-Bouldin Index. The results establish K = 3 as the outcome because it is the most pedagogically actionable, with a Silhouette Score of 0.553 and a Davies-Bouldin Index of 0.849, dividing students into High (11.1%), Medium (70.0%), and Low (19.0%) understanding groups. In conclusion, K-Means can map students' level of understanding of ICT objectively, providing a basis for teachers to design more targeted learning strategies.

Implementation of Deep Learning for Orange Variety Classification Using the Convolutional Neural Network (CNN) Method Based on Fruit Images

Iyan Abdillah Daulay, Mansur AS, Said Iskandar Al Idrus, Zulfahmi Indra — Thu, 15 Oct 2026 00:00:00 +0700

This research aims to implement the Deep Learning method in classifying visually similar types of oranges using the Convolutional Neural Network (CNN) algorithm, as well as to determine the model’s accuracy in recognizing orange fruit images. The study classifies four types of oranges: Keprok Brastepu, Keprok Maga, Murcott, and Siam Madu. The dataset used consists of 1.739 initial dataset, which increased to 2.304 images after augmentation and oversampling, with a data split ratio of 80% for training and 20% for testing. The pre-processing stages include data labeling, data organization, remove background, and image resizing to match the input requirements of the Convolutional Neural Network model. The model employed is the InceptionV3 architecture with the Adam optimizer, categorical crossentropy as the loss function, 50 epochs, and early stopping applied at the 20th epoch when accuracy reached 98%. The experimental results show that the model with a learning rate of 0,0001 achieved the best performance, with training accuracy of 0.98 and validation accuracy of 0.98, along with precision of 0.98, recall of 0.97, and an f1-score of 0.97. Meanwhile, a higher learning rate (0,01) resulted in signs of overfitting. Based on these findings, it can be concluded that the Convolutional Neural Network method with the InceptionV3 architecture is capable of classifying visually identical orange varieties with excellent accuracy, making it effective for implementation in image-based fruit identification systems.

YOLOv8-Based Real-Time Indonesian Sign Language (BISINDO) Detection System for Android Applications

Thu, 15 Oct 2026 00:00:00 +0700

The manuscript should contain an abstract. The abstract should be self-contained and citation-free and should not exceed 200 words. The abstract should state the purpose, approach, results and conclusions of the work. The author should assume that the reader has some knowledge of the subject but has not read the paper. Thus, the abstract should be intelligible and complete in it-self (no numerical references); it should not cite figures, tables, or sections of the paper. The abstract should be written using third person instead of first person. Communication barriers between Deaf individuals and people unfamiliar with Indonesian Sign Language (BISINDO) reduce the effectiveness of daily interactions. This study develops a real-time BISINDO recognition system for Android by integrating YOLOv8, MediaPipe Hands, and a one-dimensional convolutional neural network (Conv1D). YOLOv8 is used to detect the user as the region of interest, while MediaPipe Hands extracts 21 hand landmarks that are processed by the Conv1D model to classify 15 BISINDO gesture classes. A balanced dataset consisting of 900 gesture samples was used for training and evaluation. The proposed architecture contains 59,680 trainable parameters, making it lightweight and suitable for mobile deployment. Experimental results show stable convergence during training, increasing training and validation accuracy, and decreasing loss values. The classification process correctly recognizes most gesture classes, with only limited confusion among visually similar gestures. The developed Android application performs real-time gesture recognition using a smartphone camera and converts recognized gestures into text. The proposed approach demonstrates that the integration of YOLOv8, MediaPipe Hands, and Conv1D provides an efficient and practical solution for real-time BISINDO recognition and has the potential to improve communication accessibility through mobile assistive technology.

Design and Implementation of an IoT-Based Automatic Plant Watering System Using ESP32 and the Blynk Platform

Qois Ridwan, Yusuf Eko Rohmadi, Muhammad Abdul Azis, Taman Ginting — Thu, 15 Oct 2026 00:00:00 +0700

Manual watering of potted plants in household-scale gardening is often inconsistent due to limited time and high mobility of plant owners, which can cause water waste or plant stress from irregular care. This study designs and implements an automatic monitoring and control system for plant watering based on the Internet of Things (IoT), using an ESP32 microcontroller and the Blynk platform. The system uses a resistive soil moisture sensor to continuously read real-time soil conditions and applies a threshold-based approach so that irrigation is triggered only when needed, while a manual override is also available via the Blynk mobile application for flexibility. The system was developed using the Prototyping method, allowing iterative design, construction, and evaluation before final deployment. Testing covered soil moisture sensor accuracy, relay and actuator response, Wi-Fi connectivity, and automatic mode operation, followed by an overall functional test and a user satisfaction survey of 30 respondents. The results show that all hardware components and features operated according to specification, and most respondents reported satisfaction with the system's usability and reliability. The developed system offers a practical solution for household plant owners with limited time for manual watering and supports the concept of a sustainable smart home garden.

Sentiment Analysis of Mikie Holiday Funland Visitor Reviews Using the Naïve Bayes Algorithm

Lala Anjelir, Relita Buaton, Muammar Khadapi — Thu, 15 Oct 2026 00:00:00 +0700

Mikie Holiday Funland is one of the tourist destinations in North Sumatra that receives numerous visitor reviews through digital platforms such as Google Maps and TikTok. The large number of available reviews makes manual analysis less effective, creating the need for sentiment analysis to identify visitors' perceptions of the facilities and services provided. This study aims to classify visitor reviews of Mikie Holiday Funland into positive, negative, and neutral sentiment categories using the Naïve Bayes method. This research adopted the CRISP-DM framework, with data processing stages including text preprocessing, lexicon-based sentiment labeling, TF-IDF weighting, and classification using the Naïve Bayes algorithm. The data used in this study were collected from Google Maps reviews and TikTok comments from February 2025 to February 2026, totaling 4,876 reviews. Model evaluation was conducted using a confusion matrix and classification report. The results showed that the Naïve Bayes method achieved an accuracy of 73% in classifying review sentiments. Of the 4,876 reviews analyzed, 849 reviews (17.4%) were classified as positive, 1,466 reviews (30.1%) as negative, and 2,561 reviews (52.5%) as neutral. The dominance of neutral sentiment indicates that most reviews were informative and did not express strong opinions. Furthermore, the analysis results were successfully implemented in a Streamlit-based application to provide interactive sentiment visualizations. These findings are expected to serve as evaluation material for improving the quality of tourism facilities and services, as well as a source of information for prospective visitors in making decisions about visiting Mikie Holiday Funland.

Design and Implementation of an Intelligent Agent–Based Village Certificate Service System with Digital Signature Support

Thu, 15 Oct 2026 00:00:00 +0700

The reliance on manual procedures in village administrative services continues to create significant operational challenges, including slow document issuance, high rates of clerical errors, and insufficient mechanisms for verifying the authenticity of official documents. This study aimed to design and implement an Intelligent Agent–Based Village Certificate Service System with Digital Signature Support at Hilimbosi Village Office, Sitolu Ori District, North Nias Regency, North Sumatra. The system was developed by following the Waterfall software development model, using Laravel (PHP) as the backend framework, MySQL as the relational database management system, and OpenSSL/PKI for digital signature operations. A rule-based intelligent agent was integrated into the system to automate the validation of application data and to initiate document generation without requiring repeated manual intervention from clerical staff. Digital signatures based on the RSA-2048 algorithm and the SHA-256 hashing function were implemented to guarantee the authenticity and integrity of electronically issued certificates. Evaluation results demonstrated that the system reduced the average document processing time from two to three working days to under ten minutes, decreased the data error rate from 12% to below 2%, and received positive feedback from participants during User Acceptance Testing (UAT). This research contributes a replicable integration model combining intelligent agent technology and digital signature mechanisms within a village-level public administration context.

Keywords: intelligent agent, digital signature, village certificate, village information system, PKI, Laravel, Waterfall

Implementation of the MOORA Method in a Web-Based Decision Support System for Selecting Cat Food for Cats with Digestive Disorders

Anni Tias Lestari, Adam Muiz — Thu, 15 Oct 2026 00:00:00 +0700

Selecting appropriate food for cats with digestive disorders is a challenge for many pet owners due to the variety of products available with different nutritional contents, prices, and brands. Inappropriate food selection may affect the recovery process and overall health of cats. This study aims to develop a web-based Decision Support System (DSS) that assists users in selecting the most suitable cat food by implementing the Multi-Objective Optimization on the Basis of Ratio Analysis (MOORA) method. The system evaluates several criteria, including price, moisture content, protein content, fat content, and crude fiber content, to rank available alternatives objectively. The system was developed using the Personal Extreme Programming (PXP) approach and implemented as a web application to improve accessibility and usability. The results show that the MOORA method successfully provides objective recommendations based on the specified criteria and simplifies the decision-making process for users. System testing using Black Box Testing, White Box Testing, and User Acceptance Testing indicates that all system functions operate correctly and meet user requirements. Therefore, the proposed system can effectively support pet owners in selecting appropriate cat food while improving service quality at pet stores.

Design of a Website-Based Congregation Data Information System (Case Study: GKS Galutosi)

Hosea Ngunju Lenang, Yustina Rada — Thu, 15 Oct 2026 00:00:00 +0700

This study aims to design and build a website-based congregation data information system to assist the Congregation Council Executive Board (Badan Pelaksana Majelis Jemaat, BPMJ) of GKS Galutosi in managing congregation data more quickly, accurately, and efficiently. The system was developed using the Rapid Application Development (RAD) method, covering the requirements planning, design, development, and implementation stages. Built using PHP Native, MySQL, and Tailwind CSS, the system features congregation data management, commission grouping, family head data, neighborhood zones, church council data, and a comprehensive reporting module. Black Box Testing on 21 scenarios achieved a 100% success rate, and System Usability Scale (SUS) testing with 12 respondents yielded an average score of 96.88 (Grade A/Excellent), confirming a very high level of usability and user acceptance.

Naïve Bayes Method Implementation in Selecting Olympiad Participants for Grade XI Students of SMA Kristen Waibakul

Irna Winarti Rambu Lika, Fajar Hariadi — Thu, 15 Oct 2026 00:00:00 +0700

The selection of academic Olympiad participants at SMA Kristen Waibakul has traditionally relied on manual judgement of report-card scores and teacher recommendations, a practice that is prone to subjectivity and inconsistency. This study develops and empirically evaluates a decision support approach that combines the Naïve Bayes classifier with a Greedy Allocation strategy to select eleventh-grade students for a school Olympiad team spanning nine academic fields: Mathematics, Physics, Economics, Chemistry, Biology, Geography, Informatics, Earth Science, and Astronomy. A training set of 80 students and an independent test set of 40 students, each described by seven subject scores, were discretised into High/Low categories using an 80-point threshold and modelled with Laplace-smoothed prior and likelihood probabilities. Nine independent binary Naïve Bayes classifiers were trained, one per field, and their posterior probabilities were subsequently passed to a Greedy Allocation procedure that enforces a fixed quota of two students per field. Evaluated against real ground-truth labels, the combined confusion matrix over all nine fields (360 decisions) yielded an overall accuracy of 90.28%, a macro-average precision of 84%, a macro-average recall of 65%, and a macro-average F1-score of 70%. The Greedy Allocation stage produced a final roster of 18 students, of which 14 (77.78%) matched the actual selection outcome. The findings indicate that the proposed system can support faster, more consistent, and auditable Olympiad selection, although recall for the minority "pass" class remains a limitation that warrants further attention.

Development Of A Mobile-Based Digital Student Id Card Application Using The Flutter Framework

Wahit Fitriyanto, Muhammad Abdul Aziz, Yusuf Eko Rohmadi, Taman Ginting — Thu, 15 Oct 2026 00:00:00 +0700

The management of physical Student ID Cards at Boyolali University has faced printing inefficiencies and the risk of forgery due to the rapidly growing number of students over the past five years. This research aims to develop a mobile-based Digital KTM application using the Flutter framework and Appwrite backend to provide a secure and efficient identity solution. The software development method applied is the Waterfall model, which includes the stages of requirements analysis, UML diagram, ERD, code construction, and deployment. System security is enhanced through the integration of local fingerprint biometric authentication based on the BiometricPrompt API connected to a Trusted Execution Environment, as well as the implementation of a dynamic QR code based on SHA-256 cryptography with a five-minute expiration limit to prevent visual duplication. Functional testing results using Black Box Testing on 13 test case scenarios showed a high success rate. Usability evaluation via a Likert scale questionnaire administered to 30 respondents yielded an average usability score of 83.70%, indicating that this application is highly suitable for full implementation. This research successfully eliminated the weaknesses of physical cards, reduced issuance time from weeks to minutes, and supports the realization of the Smart Campus vision at Boyolali University.

Design and Implementation of an IoT-Based Automatic Salted Fish Dryer System Using Load Cell as an Objective Moisture Indicator

Darmiati, Abdullah Hanif, Agus Triyono — Thu, 15 Oct 2026 00:00:00 +0700

Fish is a highly nutritious but rapidly perishable food commodity. Traditional salted fish drying methods heavily depend on unpredictable weather conditions, often taking days to complete and leaving the product prone to contamination. While automated dryers have been introduced, most lack an objective metric to halt the process, relying instead on subjective visual inspection. This study aims to design and implement an Internet of Things (IoT)-based automatic salted fish dryer system equipped with a load cell to objectively determine the dryness level. The proposed system utilizes an ESP32 microcontroller, a DHT22 sensor, a PTC heater, and a DC fan to automatically maintain the drying chamber at an optimal temperature of 70°C. Furthermore, a load cell sensor continuously monitors weight reduction, serving as a real-time proxy for the moisture content. Operational data is transmitted via Wi-Fi to the ThingsBoard platform for remote visualization. System validation showed that the DHT22 and load cell sensors have an accuracy of over 99.20%. Experimental results on 1 kg of Sepat and Layang fish demonstrate that the automated system significantly accelerates the drying process, reducing the required time from 2-3 days to just 8 hours. The system successfully executed an automatic cut-off when the fish weight dropped by 50%, ensuring hygienic, efficient, and consistent production for small-to-medium enterprises.

Design and Construction of a Storage Box with Facial Recognition for Authentication

Andika Julian Pratama, Abdullah Hanif, Irwansyah — Thu, 15 Oct 2026 00:00:00 +0700

Conventional storage box systems still rely on physical keys that are prone to being lost. This study aims to design a smart storage box system based on facial recognition to enhance security and efficiency. The approach utilizes an ESP32-CAM module integrated with the Haar Cascade method for real-time face detection. The system's operation involves stages of face detection, feature extraction, and database matching. Upon successful authentication, the solenoid lock opens automatically; conversely, access is denied if the face is unrecognized. Testing on 30 facial image samples demonstrated a 100% success rate under normal lighting conditions. Additionally, the system recorded an average response time of one second for the lock to open. It can be concluded that the system operates stably and effectively as a keyless security alternative for storage boxes.

Sentiment Analysis of Public Opinion on the Whoosh High-Speed Railway Debt Controversy on Facebook Using the Naïve Bayes Model

Meisyari Andini, Akim Manaor Hara Pardede, Muammar Khadapi — Thu, 15 Oct 2026 00:00:00 +0700

The controversy surrounding the Jakarta-Bandung High-Speed Rail (Whoosh) project’s debt has triggered significant public opinion polarization on Facebook. This research aims to analyze public sentiment regarding this financial issue to transform unstructured data into strategic insights for stakeholders. The applied methodology follows the CRISP-DM framework using the Naïve Bayes Classifier algorithm reinforced by Term Frequency-Inverse Document Frequency (TF-IDF) weighting. Primary data consisted of 2,000 comments from the "Nusa Update" fan page, which underwent thorough text preprocessing including cleansing, normalization, and stemming to reduce noise. Results indicate that the classification model achieved an overall accuracy of 83.75%, exceeding the 80% target. Sentiment distribution analysis reveals a dominance of negative opinions at over 80%, reflecting public concern over cost overruns and the burden on the state budget (APBN). These findings are implemented into an interactive Streamlit-based dashboard to facilitate real-time monitoring of public policy reputation. This study confirms the effectiveness of hybrid approaches in transforming "digital waste" into measurable scientific information.

Implementation of a Web-Based Decision Support System for Employee Performance Evaluation Using the TOPSIS Method at Cinta Quran Center

Muhammad Alfian Ridho Donafan, Resti Amalia — Thu, 15 Oct 2026 00:00:00 +0700

Employee performance evaluation is an essential process for improving human resource quality and supporting decision-making within an organization. At Cinta Quran Center, the employee performance evaluation process is still conducted manually, resulting in a time-consuming process and increasing the risk of data processing errors. This study aims to develop a web-based Decision Support System using the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method to support a more objective and efficient employee performance evaluation process. The system was developed using the Prototype development method with the Laravel framework, PHP programming language, and MySQL database. The results demonstrate that the proposed system is capable of managing employee assessment data, performing TOPSIS calculations automatically, generating employee rankings based on preference values, and producing evaluation reports quickly and accurately. Furthermore, Black Box Testing confirmed that all system functionalities operated as expected, indicating that the developed system effectively supports employee performance evaluation at Cinta Quran Center.

Implementation of Firewall and Internet Bandwidth Management for Civil Servants at the BKPSDM of Central Sumba Regency

Didi Napoleon Umbu Tara Dima, Pingky Alfa Ray Leo Lede — Thu, 15 Oct 2026 00:00:00 +0700

Advancements in information technology require government agencies to maintain secure and stable computer network infrastructures to support operational efficiency. The BKPSDM (Personnel and Human Resources Development Agency) of Central Sumba Regency relies heavily on internet connectivity to manage State Civil Apparatus (ASN) personnel data. However, staff survey results indicate that current internet quality is suboptimal, characterized by frequent slowness and instability caused by the use of bandwidth for non-productive activities, specifically accessing entertainment platforms like YouTube and TikTok. Furthermore, the office WiFi network is vulnerable to unauthorized access, posing a threat to the confidentiality of ASN data. This study aims to implement a network security and bandwidth management system using MikroTik Routerboard devices, grounded in Government Regulation (PP) Number 94 concerning Civil Servant Discipline. The research methodology encompasses network analysis, topology design, implementation, and testing. Implementation results demonstrate that TLS Host-based firewall rules successfully blocked all access to YouTube and TikTok within the ASN work segment. Bandwidth management utilizing the "Simple Queue" feature with a hierarchical method effectively segregated data traffic between the ASN segment and the Public WiFi segment. Quality of Service (QoS) testing revealed significant improvements in network performance: average latency dropped from 185 ms to 32 ms, packet loss decreased from 8.4% to 0.12%, and throughput stabilized at 34.1 Mbps within the ASN segment. This study confirms that integrating firewall capabilities and bandwidth management can optimize data security, stabilize internet connections, and support the enforcement of ASN working-hour discipline at the BKPSDM of Central Sumba Regency.

Clustering of Student Complaints Against the Implementation of Full Day School Using the K-Means Algorithm

Nisa Ramadani, Suci Ramadani, Juliana Naftali Sitompul — Thu, 15 Oct 2026 00:00:00 +0700

The full-day school policy extends the duration of learning activities at school with the aim of improving the quality of education and character building of students. In practice, the addition of study time does not always run in line with student readiness, thus giving rise to various complaints related to fatigue, difficulty concentrating, excessive assignment burden, and lack of rest time. The diversity of these complaints is complex and cannot be analyzed subjectively, so a structured approach is needed to identify the dominant patterns of complaints objectively. This study applies the data mining approach through the k-means clustering algorithm to group student complaints against the implementation of full-day school. The data used are 523 records obtained from a questionnaire, with three variables, namely class, complaint, and student suggestion, which are then transformed into numerical values and processed using MATLAB R2014a. The grouping is tested with three configurations, namely 3, 4, and 5 clusters, and the cluster quality is evaluated based on the cluster variance value. The results show that the configuration of 5 clusters produces the lowest cluster variance of 20.525 compared to 3 clusters (40.767) and 4 clusters (31.681), so that 5 clusters is the most optimal and representative configuration. The type of complaint is the most dominant factor in shaping the grouping pattern. These findings can be used by the school as a basis for evaluating the implementation of full-day school in a more targeted manner in accordance with the characteristics of student complaints.

Early Flood Warning System in Samarinda City

Gusvi Ainur Ikhsan Isnor, Abdullah Hanif, Irwansyah — Thu, 15 Oct 2026 00:00:00 +0700

This study aims to design and implement an Early Flood Warning System in Samarinda City to help the community monitor potential flooding in real-time. The system uses an ESP32 microcontroller connected to the Firebase Realtime Database and is equipped with two main sensors: a raindrop sensor to detect rainfall intensity and an ultrasonic sensor to measure water surface height. The sensor data are displayed on an LCD 16x2 and sent directly to Firebase, which can then be accessed via an Android application or a local web interface based on HTML. The system also includes a buzzer that serves as an alarm when the detected condition reaches a danger level. In addition, a fuzzy logic method is applied to determine the flood risk level based on the combination of rainfall intensity and water surface distance. Based on the test results, the system operates stably with a high level of 80% accuracy in displaying data and delivering timely warning notifications. Therefore, this system can serve as an effective and practical flood mitigation tool for communities in the Samarinda area.

Design and Construction of an Information System for Monitoring Vegetation Density and Condition Based on Google Earth Engine using Multi Spectral Vegetation Indices of Landsat 8 Collection 2 Imagery in Surabaya

Thu, 15 Oct 2026 00:00:00 +0700

Green open spaces play an important role ecologically, socially, aesthetically, and economically in supporting the quality of the urban environment; however, monitoring their condition in the City of Surabaya still requires an efficient, spatial-data-driven approach. This study aims to design and develop a vegetation monitoring information system based on Google Earth Engine (GEE) capable of identifying the role and influence of vegetation indices on urban environmental conditions. The system was developed using the Waterfall method, leveraging Landsat 8 Collection 2 satellite imagery from the 2020–2025 period, which was processed through six spectral vegetation indices NDVI, EVI, MSAVI2, NDMI, GNDVI, and NBR to measure vegetation density, chlorophyll content, canopy moisture, and vegetation stress levels

Comparison of Decision Tree and Random Forest Algorithms for Diabetes Classification

Thu, 15 Oct 2026 00:00:00 +0700

Diabetes merupakan salah satu penyakit kronis yang paling umum di seluruh dunia, dengan jumlah penderita yang terus meningkat setiap tahunnya. Perkembangan pesat Machine Learning telah menyediakan pendekatan efektif untuk meningkatkan akurasi klasifikasi diabetes. Studi ini bertujuan untuk membandingkan kinerja algoritma Decision Tree dan Random Forest dalam mengklasifikasikan diabetes menggunakan Dataset Diabetes Suku Indian Pima. Dataset ini terdiri dari 768 catatan pasien dengan delapan atribut prediktor dan satu atribut target. Metodologi penelitian meliputi pra-pemrosesan data, pembagian dataset menjadi set pelatihan dan pengujian, pengembangan model, dan evaluasi kinerja menggunakan matriks kebingungan, akurasi, presisi, recall, dan F1-score. Hasil eksperimen diharapkan dapat mengidentifikasi algoritma dengan kinerja klasifikasi terbaik dan memberikan referensi berharga untuk menerapkan teknik Machine Learning di sektor kesehatan.

Implementation of a Mobile Application for Al-Qur'an Memorization Evaluation Using a Gamification Approach and the SDLC Method

Rafly Putra, Budi Apriyanto — Thu, 15 Oct 2026 00:00:00 +0700

The Quran memorization evaluation process at Majelis Taklim Wasilatul Muhtadin is currently conducted manually; consequently, the recording of recitation submissions, the assessment process, and the monitoring of student progress are not operating optimally. This study aims to develop an Android-based mobile application utilizing a gamification approach to support a more effective and structured memorization evaluation process. The application was developed using the Software Development Life Cycle (SDLC) methodology with a Prototyping model. It was built in Android Studio using the Kotlin programming language and incorporates Firebase Authentication and Firebase Firestore. Key features include user login, memorization submission, memorization validation, practice sessions, quizzes, and exam package management. Gamification elements were implemented through features such as levels, memorization progress tracking, daily missions, quizzes, and exams. Testing results indicate that all application functions operate according to user requirements, achieving a user acceptance rate of 91.18% based on questionnaire data. Thus, the application serves as a supportive tool for Quran memorization evaluation while enabling instructors to monitor student progress more effectively.

Performance Comparison of Random Forest and XGBoost Algorithms for Used Car Price Prediction Using a Kaggle Dataset

Thu, 15 Oct 2026 00:00:00 +0700

Accurate prediction of used car prices plays an essential role in supporting fair vehicle valuation and informed decision-making within the automotive market. This study aims to compare the performance of the Random Forest and XGBoost algorithms for predicting used car prices using a publicly available Kaggle dataset. The proposed methodology includes data collection, data preprocessing, feature encoding, model training, and model evaluation. The dataset is divided into training and testing subsets to ensure an objective assessment of predictive performance. Both algorithms are evaluated using Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and the Coefficient of Determination (R² Score). The comparative analysis is intended to identify the algorithm that provides superior prediction accuracy and better generalization capability. The findings of this research are expected to contribute to the development of reliable machine learning models for used car price prediction and provide valuable insights for buyers, sellers, and automotive businesses.

Ear Disease Classification System Using Convolutional Neural Network with MobileNetV2 Architecture

Ahmad Zakiyyun Ibadurrahman, Yisti Vita Via, Achmad Junaidi — Thu, 15 Oct 2026 00:00:00 +0700

Ear disease can affect people of all ages and, if not detected early, may lead to complications such as permanent hearing loss. Limited access to early screening remains a major obstacle to ear disease detection in Indonesia. This study aims to develop a system that classifies tympanic membrane images into two categories, normal and abnormal, using a Convolutional Neural Network (CNN) with the MobileNetV2 architecture and a transfer learning approach. The dataset was obtained from Kaggle with 956 images, reduced from nine disease classes into two classes and expanded to 5,736 images through flip and rotation augmentation. Six testing scenarios were designed by varying the epoch (10, 20, 30) and batch size (16, 32) to observe the effect of both hyperparameters on model performance. Testing on 861 test images produced accuracies ranging from 79.4% to 83.1%. The scenario with 30 epochs and a batch size of 16 achieved the highest accuracy (83.1%) and precision (85.4%), while the scenario with 10 epochs and a batch size of 16 achieved the highest recall (90.4%) with the fastest training time (18 minutes). A trade-off pattern between precision and recall was observed, in which models trained with fewer epochs tended to be more aggressive in detecting the abnormal class. For early screening purposes that prioritize sensitivity, the scenario with 10 epochs and a batch size of 16 is recommended because it minimizes false negative cases.

Classification of Banana Ripeness Using Transfer Learning VGG16 and Logistic Regression

Salma Julfa Iswanto , Robi Aziz Zuama, Arina Selawati — Thu, 15 Oct 2026 00:00:00 +0700

Banana ripeness classification plays an important role in improving the efficiency of postharvest handling, sorting, and distribution processes. Conventional visual inspection is subjective and often produces inconsistent results due to human judgment. This study proposes an automatic banana ripeness classification model by combining Transfer Learning VGG16 as a feature extractor with Logistic Regression as the classification algorithm. The dataset consists of 1,490 images of Cavendish bananas categorized into four ripeness levels: unripe, semi-ripe, ripe, and overripe. Image feature extraction was performed using the pre-trained VGG16 model, while Logistic Regression was employed to classify the extracted feature vectors. The proposed model was implemented using Orange Data Mining and evaluated under three validation scenarios, namely 10-fold cross-validation, 80:20 random sampling, and 70:30 random sampling. Model performance was assessed using Accuracy, Precision, Recall, F1-Score, Area Under the Curve (AUC), Matthews Correlation Coefficient (MCC), Confusion Matrix, and ROC Curve. Experimental results indicate that the 10-fold cross-validation scenario achieved the highest performance, obtaining 97.8% accuracy, 97.8% precision, 97.8% recall, 97.8% F1-score, 0.999 AUC, and 0.970 MCC. These findings demonstrate that the proposed approach provides accurate and computationally efficient banana ripeness classification, making it suitable for supporting automated fruit sorting applications.

Systematic Evaluation of YOLOv8 Training Configurations for Detecting Bovine Sperm Head Abnormalities

Rizqy Khoirul Waritsin, I Gede Susrama Mas Diyasa , Eva Yulia Puspaningrum — Thu, 15 Oct 2026 00:00:00 +0700

Bovine sperm morphology assessment is an important stage in evaluating semen quality for artificial insemination programs. Manual examination under a microscope is time-consuming and highly dependent on the observer's expertise, motivating the development of automated detection systems using deep learning. This study systematically evaluates the influence of YOLOv8 training configurations for detecting bovine sperm head abnormalities. A total of 27 experimental scenarios were designed by combining three optimizers (SGD, Adam, and RMSProp), three learning rates (0.0001, 0.001, and 0.01), and three training epochs (75, 100, and 125). Contrast Limited Adaptive Histogram Equalization (CLAHE) was applied to improve image quality before model training. Performance was evaluated using Precision, Recall, F1-score, and mAP@0.5. The experimental results indicate that the training configuration significantly affects detection performance. The best performance was achieved using the RMSProp optimizer with a learning rate of 0.01 and 75 training epochs, producing a Precision of 89.81%, Recall of 88.38%, F1-score of 89.08%, and mAP@0.5 of 91.76%. These findings demonstrate that systematic evaluation of training configurations provides practical guidance for improving YOLOv8-based bovine sperm abnormality detection without modifying the network architecture

Analysis and Implementation of Network Security System Using MikroTik Firewall and Web-Based IDS with NDLC Method at Rose Madaniyah Residential Area

Muhammad Aqil Farhan, Herwis Gultom — Thu, 15 Oct 2026 00:00:00 +0700

The use of community-based internet networks (RT/RW Net) in the Rose Madaniyah residential area is currently vulnerable to facility misuse and cyber threats due to the lack of centralized security and logging systems. This study aims to analyze and implement a multi-layered network security system as a preventive measure using a MikroTik router and an Intrusion Detection System (IDS). The infrastructure development is guided by the Network Development Life Cycle (NDLC) method, where firewall filter rules on MikroTik are implemented to filter basic access, while the Snort IDS sensor is tasked with deeply detecting suspicious activity patterns. To facilitate monitoring without requiring command-line expertise, this defense system is integrated with a web-based monitoring dashboard application using PHP and MySQL. The implementation results show that the residential network infrastructure has become more secure and proactive. The monitoring dashboard is proven capable of visualizing alert log data from the Snort sensor in real-time, enabling community management to monitor network status and take preventive actions against potential disruptions.

Implementation of Machine Learning using the Multiple Linear Regression Method for Boarding House (Kost) Rental Price Prediction Around Universitas BSI Jatiwaringin Campus

David Utomo, Ahmad Fauzi, Andi Sanjaya — Thu, 15 Oct 2026 00:00:00 +0700

SIPREKOS is a web-based boarding house (kost) rental price prediction system developed to help students and kost owners around Universitas BSI Jatiwaringin Campus obtain objective price estimates. High rental prices and limited price transparency motivate the need for such a system. This study builds a prediction model using the Multiple Linear Regression method with polynomial features, based on nine kost predictor variables together with a squared-distance derivative variable to capture the non-linear effect of distance on price. Primary data were collected manually from 115 kost units listed on Mamikos.com within a 3 km radius of the campus during May–June 2026, then cleaned using an absolute-error-based residual analysis method, resulting in 100 valid records. The valid data were split 80:20 into training and test sets; regression coefficients were estimated from the training set using the Normal Equation. Testing shows the model explains 72.4% of rental-price variance (R² = 0.724) with an average prediction error of IDR 116,969 (MAE), and it passes the classical assumption tests (multicollinearity, heteroscedasticity, and normality), confirming that the estimator is BLUE (Best Linear Unbiased Estimator). Black-box testing across ten scenarios shows that all system functions, for both general users and administrators, operate according to specification.

Analysis of Public Sentiment Toward Government Digitalization Using NLP and Deep Learning in Python

Muhammad Ibnu Ramadhan — Thu, 15 Oct 2026 00:00:00 +0700

This study aims to analyze public sentiment toward the Indonesian government’s digitalization policies on the X social media platform using Natural Language Processing (NLP) and Python-based deep learning approaches. The research data consist of posts related to the Electronic-Based Government System (SPBE), One Data Indonesia, and Digital Identity policies collected during 2025–2026. The research stages include data collection, text preprocessing, non-organic account filtering, sentiment labeling, model development using deep learning algorithms (LSTM, CNN, and IndoBERT/IndoBERTweet), and performance evaluation using accuracy, precision, recall, and F1-score metrics. Furthermore, topic analysis is conducted to identify policy issues that generate the highest proportion of negative sentiment. The expected outcomes are an accurate sentiment classification model and a comprehensive understanding of public perceptions of governmental digital transformation. The findings are anticipated to provide a foundation for government agencies to formulate communication strategies and improve digital public services in a manner that is more responsive to citizens’ needs.

Monitoring System for a Chili Plant Pest Detection and Repellent Device Using Liquid Pesticides Based on the Internet of Things (IoT)

Pia Dwi Windiani, Relita Buaton, Milli Alfhi Syari — Thu, 15 Oct 2026 00:00:00 +0700

Pest attacks, along with changes in environmental temperature and humidity, are factors that affect the growth and productivity of chili plants. Conventional monitoring methods are considered less effective because they require significant time and direct human involvement. This study aims to design and implement an Internet of Things (IoT)-based monitoring, pest detection, and pest repellent system for chili plants using the ESP32 microcontroller. The system integrates a DHT22 sensor to monitor temperature and humidity, a Passive Infrared (PIR) sensor to detect movement around the plants, a relay and a DC pump to control automatic pesticide spraying, a buzzer as an audible notification, and an ESP32-CAM for real-time visual monitoring. Sensor data are displayed in real time through the Blynk application, allowing users to remotely monitor plant conditions. The testing results indicate that the system is capable of monitoring temperature and humidity, detecting movement, automatically activating the buzzer and DC pump, and displaying real-time visual conditions of the plants through the ESP32-CAM. Therefore, the proposed system is capable of supporting real-time monitoring and pest control for chili plants based on the Internet of Things.

User Satisfaction Analysis of the E-Berpadu Website at the Bekasi District Court Using the C4.5 Algorithm

Muhammad Fikri Hidayat, Eva Rahmawati, Sri Diantika — Thu, 15 Oct 2026 00:00:00 +0700

The development of electronic government (e-government) has encouraged public institutions to improve the quality of digital services to provide more effective, transparent, and user-oriented services. One implementation is the e-Berpadu (Integrated Electronic Criminal Case Files) website developed by the Supreme Court of the Republic of Indonesia to support electronic criminal case administration. This study aims to measure user satisfaction with the e-Berpadu website at the Bekasi District Court, develop a user satisfaction classification model using the C4.5 algorithm, and identify the factors influencing user satisfaction. The study employed a quantitative method with a data mining approach involving 265 respondents. The research variables consisted of System Quality, Information Quality, Service Quality, and System Use. The research process included data cleaning, transformation, labeling, data selection, an 80:20 data split, C4.5 model development, and performance evaluation using accuracy, precision, recall, and F1-score. The results showed that 96.23% of respondents were satisfied, while 3.77% were dissatisfied. The pruned C4.5 model achieved an accuracy of 98.11%, precision of 98.15%, recall of 98.11%, and an F1-score of 97.81%. Information Quality, particularly the clarity of feature usage instructions (IQ2), was identified as the most influential factor, followed by System Use (SU2), Service Quality (SV2), and System Quality (SQ2). These findings provide valuable insights for improving the quality of the e-Berpadu website.

Implementation of The K-Means Clustering Algorithm for Grouping Oil Palm Land Productivity Levels (Case Study: PTPN IV Regional II, Tanjung Jati Plantation)

Kesya Alya — Thu, 15 Oct 2026 00:00:00 +0700

PTPN IV Regional II Tanjung Jati Estate is a state-owned enterprise engaged in the agribusiness sector, particularly in oil palm plantation management, which is its primary commodity. This study aims to apply the K-Means Clustering algorithm to group oil palm plantation land based on plant age, land area, and production yield. The results show that the plantation land data were successfully grouped into three clusters with different characteristics. Cluster 0 has an average plant age of 10 years, a land area of 7.90 hectares, and a production yield of 174.67 tons per year. Cluster 1 has an average plant age of 22 years, a land area of 17.26 hectares, and a production yield of 347.76 tons per year. Cluster 2 has an average plant age of 9 years, a land area of 21.81 hectares, and a production yield of 481.97 tons per year. The best clustering result was obtained at K = 3 with a Davies-Bouldin Index (DBI) value of 0.241. The clustering results provide information regarding the characteristics of each land group, which can assist the company in understanding the condition of the managed plantation land and support the evaluation process of plantation land data.

Security Testing of the Systematic UNISI Website using the Penetration Testing Method

Agung Ramadhan, Abdul Muni, Bayu Rianto — Thu, 15 Oct 2026 00:00:00 +0700

Academic websites require continuous security evaluation because they support information publication and institutional services that are accessible through the internet. This study evaluates the security of the Sistemasi UNISI website using a non-destructive penetration testing method with OWASP Zed Attack Proxy (ZAP). The research stages consisted of literature study, planning, reconnaissance, scanning, vulnerability testing, report analysis, conclusion, and remediation formulation. Data were collected through literature review and observation of the target website, while testing was performed by entering the Sistemasi UNISI URL into the Automated Scan feature of OWASP ZAP. The detailed scan report identified 19 findings consisting of four medium-risk findings, eight low-risk findings, and seven informational findings, with no high-risk finding reported. The medium findings were absence of Anti-CSRF tokens, Content Security Policy header not set, missing anti-clickjacking header, and vulnerable JavaScript library. The results indicate that the website still requires security hardening, especially in browser security headers, cookie configuration, JavaScript dependency management, HTTPS consistency, and information disclosure control. The recommended improvements can reduce exploitation opportunities and strengthen the security posture of the website

Implementation of the Topsis Method in a Digital Book Collection Management Information System to Support School Digital Transformation: A Case Study of SMK Nusa Bangsa

Fitra Fitra Raihan, Resti Amalia — Thu, 15 Oct 2026 00:00:00 +0700

Libraries play an important role in supporting educational activities and providing access to information. As technology continues to advance, library management needs to transform into a digital system to make services more effective and accessible. Based on observations conducted at SMK Nusa Bangsa, several challenges were identified, including slow book borrowing processes, difficulties in monitoring book collections, delays in book returns, and low student interest in visiting the library. This study developed a web-based digital book collection management information system by implementing the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method to assist in determining priorities for book collection management. The implementation results indicate that the system improves the efficiency of library management, accelerates the borrowing process, and facilitates students and teachers in accessing information about book collections. Therefore, the proposed system is expected to support the school's digital transformation and enhance the quality of library services at SMK Nusa Bangsa.

IoT-Based Automatic Monitoring and Watering System for California Lemon Tree Seedlings with Web Dashboard and Telegram Bot Integration

Muhamad Aldo Ryanda, Ahmad Rofiq Hakim, Fara Triadi — Thu, 15 Oct 2026 00:00:00 +0700

Manual watering of California lemon (Citrus limon L.) seedlings often leads to inconsistent soil moisture, which can cause root rot from overwatering or plant stress from drought [1]. This study presents the design and implementation of an Internet of Things (IoT)-based automatic watering system integrated with a web dashboard and Telegram bot. The system employs an ESP32 microcontroller connected to a soil moisture sensor (YL-69) for detecting soil water content and an ultrasonic sensor (HC-SR04) for monitoring water tank capacity. Sensor data is transmitted to Firebase Realtime Database and displayed on a web dashboard, while a Telegram bot named "AutoLemon" provides remote notifications and manual control capabilities [2]. The development followed a Prototype methodology encompassing communication, rapid design, prototype construction, and user feedback. Field testing demonstrated that the system operates as designed: the water pump activates automatically when soil moisture drops below 40% and deactivates once moisture reaches the 50–75% range. Compared to manual watering, which consumed an average of 600 ml per day, the automated system reduced water usage to approximately 479 ml per day—a savings of roughly 20%. The integration of dual monitoring interfaces enables real-time oversight of both soil conditions and water availability, offering a practical solution for small-scale lemon cultivation.

Classification of Employee Performance Based on Evaluation Data Using the K-Nearest Neighbor (K-NN) Method: (A Case Study of PTPN IV Regional II Tanjung Jati Estate)

Ayu Nabila — Thu, 15 Oct 2026 00:00:00 +0700

Employee performance evaluation is an important aspect of supporting organizational decision-making to improve productivity and work effectiveness. At PTPN IV Regional II, the performance evaluation process for fixed-term contract employees PKWT is still carried out manually based on attendance, production, and achievement data, making the process relatively time-consuming and increasing the possibility of errors in determining employee performance categories. Therefore, this study aims to apply the K-Nearest Neighbor (K-NN) algorithm to classify employee performance by utilizing RapidMiner as a tool for data processing and model evaluation. Before the classification process, the data were normalized using the Min-Max Normalization method to ensure that all attributes had the same value range. The training and testing processes were then performed using the K-Nearest Neighbor algorithm based on Euclidean Distance calculations. This study employed attendance, production, and achievement as the classification attributes to categorize employee performance into Excellent, Good, Fair, and Poor. The experimental results showed that the proposed classification model achieved an accuracy of 90.32%, indicating that the K-Nearest Neighbor algorithm can provide reliable classification results and effectively support employee performance evaluation in a faster, more objective, consistent, and accurate manner.

Design of an Interactive Multimedia Simulation Game for Chemistry Learning (Case Study: SMA Negeri 3 Waingapu)

Natalia Susana Lero, Arini Aha Pekuwali, Trisari Dewi Novyanti Bertha Mira — Thu, 15 Oct 2026 00:00:00 +0700

Chemistry learning in senior high schools is often considered difficult because many topics are abstract in nature, such as chemical reactions and the fundamental laws of chemistry. Based on the results of observations and interviews conducted at SMA Negeri 3 Waingapu, the learning process is still dominated by the lecture method, and laboratory practical activities are limited. As a result, many students experience difficulties in understanding the subject matter. Of the 36 students, approximately 65% had not achieved the Learning Objectives Achievement Indicator (IKTP). Therefore, a more interactive and engaging learning medium is needed to help students understand chemistry concepts more effectively. This study aims to design and develop an interactive multimedia simulation game as a learning medium for chemistry, implement the Decision Tree algorithm within the game system, and improve students' understanding of chemistry concepts. The expected benefits of this study are to provide a more engaging learning medium for students and to assist teachers in delivering chemistry lessons more effectively. The development method used in this research is the Multimedia Development Life Cycle (MDLC), which consists of six stages: concept, design, material collecting, assembly, testing, and distribution. Data were collected through observations, interviews, and questionnaires. The application was developed using Unity with the C# programming language and deployed on the Android platform. System testing was carried out using the Black Box Testing method. The proposed solution is an interactive multimedia simulation game that provides learning materials, chemical reaction simulations, and interactive quizzes. The system inputs consist of menu selections and quiz answers provided by users. The system processes include presenting learning materials, running chemical reaction simulations, and evaluating quiz answers using the Decision Tree algorithm. The outputs generated include quiz scores, learning evaluation results, and feedback to students regarding their learning performance.

Implementation of Yolo26 Nano Model on Riceguard AI System for Rice Leaf Disease Detection

Gilang Septia Firmansyah, Alif Bahktiar Akbar, Mufti Ari Bianto, Khairul Umam — Thu, 15 Oct 2026 00:00:00 +0700

RiceGuard AI is a new tool for diagnosing leaf diseases affecting rice plants. Early diagnosis of these diseases is crucial for maintaining rice productivity and global food security. This newly developed tool is designed to assist in diagnosing rice diseases by employing computer-vision techniques to automatically detect the condition of rice leaves. The Ultralytics YOLO26 Nano model was selected for this task because it adopts a novel "edge-first" architecture, enabling high processing speeds by eliminating the Non-Maximum Suppression (NMS) step during processing. To achieve this, the model was trained for 50 epochs, attaining a mean Average Precision (mAP) of 0.651 and an inference speed of 7.1 ms per image on an NVIDIA GTX 1650 GPU. The model achieved high accuracy for the leaf scald (99.4%) and healthy leaf (96.7%) categories but struggled to detect the narrow brown spot class (30.1%) due to visual ambiguities associated with it. YOLO26 Nano provides an accurate and efficient solution for detecting these rice conditions, enabling farmers to perform image-based diagnoses in the field for the early detection of leaf scald or healthy leaves, thereby facilitating prompt action. The model is integrated into a modern Flask web application featuring a "Glassmorphism" design style.

Analysis of Age Patterns and Factors Causing Marriage Dispensation Using the K-Means Algorithm (Case Study: Stabat Religious Court)

Della Yulia Citra, Achmad Fauzi, Ratih Puspadini — Thu, 15 Oct 2026 00:00:00 +0700

The Stabat Religious Court faces challenges in grouping the annual data on marriage dispensation applications. This data has not been optimally utilized to generate information that can support decision-making in understanding age patterns, causal factors, and areas of residence with the highest application rates. This study aims to analyze age patterns and causal factors of marriage dispensation using data mining methods with the K-Means Cluster algorithm. Determining the optimal number of clusters was carried out using the Elbow method through analysis of the Sum Of Squared Error (SEE) value, which indicated the optimal number of clusters was 3 clusters. A total of 749 marriage dispensation application data from 2021–2025 were analyzed using the Knowledge Discovery in Database (KDD) stages, which include data selection, data preprocessing, transformation, data mining (cluster), and evaluation with the help of MATLAB. The variables used in this study include the applicant's age, the reason for the marriage dispensation application, and the applicant's area of residence. The results of this study indicate that the k-means algorithm using the elbow method is capable of systematically grouping marriage dispensation data based on similar characteristics. The clustering results provide information on age patterns, causal factors, and domiciles of marriage dispensation applicants, thus providing recommendations for stabat religious courts in formulating policies and efforts to prevent early marriage.

Development of an Augmented Reality-Based Virtual Try-On Dress System for Promoting East Sumba Ikat Weaving

Brigitha Sandraningsih Regina Pawa, Yustina Rada, Eben Panja — Thu, 15 Oct 2026 00:00:00 +0700

Advancements in information and communication technology are driving digital transformation across various sectors, including the fashion industry and the promotion of local culture. One such emerging technology is Augmented Reality (AR), which enables interactive virtual visualizations. When ordering dresses made from East Sumba (traditional woven fabric), consumers often struggle to visualize how a specific dress style will pair with a particular fabric pattern prior to the actual tailoring process. Furthermore, current promotional methods for remain conventional, failing to effectively capture public interest particularly among the younger generation. This study aims to design an Augmented Reality-based virtual dress try-on information system to serve as a promotional tool for East Sumba. The research employs the Research and Development (R&D) method, comprising stages such as data collection, system requirements analysis, system design, application development, and system testing. The designed system enables the virtual visualization of dresses on mobile devices, allowing users to realistically view the combination of dress styles and fabric patterns before the tailoring process begins.

Design and Development of an Android-Based Augmented Reality Application for Simulating Men's Hairstyles Based on Face Shape

Elisabeth Rosario Viona Ndena, Pingky Alfa Ray Leo Lede — Thu, 15 Oct 2026 00:00:00 +0700

Selecting a hairstyle without considering face shape, combined with the limitations of two-dimensional reference materials, can make it difficult for customers to visualize the final result. This study aims to design and develop an Android-based Augmented Reality application for Grezfa Barbershop that interactively simulates men's hairstyles based on face shape. The research employs the Research and Development method using the Borg and Gall model, encompassing stages such as information gathering, planning, initial product development, testing, revision, and finally, dissemination and limited implementation. Data were collected through observation, interviews, questionnaires, documentation, and literature review. The application was developed using Unity 3D and Blender, while the Viola-Jones method was utilized to detect the user's facial area via the Android device's camera. The system is designed to classify face shapes into four categories: oval, round, square, and oblong. Each category offers three hairstyle options, resulting in a total of twelve low-poly 3D hairstyle models within the application. Planned testing includes functional application testing; detection testing across variations in lighting, distance, face angle, accessories, and movement; and usability assessment using the System Usability Scale with 30 male customers. The expected outcome is an application capable of providing stable, interactive, and user-friendly hairstyle simulations. The application is expected to assist customers in selecting more suitable hairstyles, minimize the gap between expectations and actual results, enhance the quality of consultations between customers and barbers, and support digital service innovation at Grezfa Barbershop. Beyond its practical benefits, this research is also intended to serve as a reference for the development of interactive visual applications based on computer vision and Augmented Reality for similar services.

Delay Prediction in BPJS Claim Payments Using a JST Backpropagation Neural Network

Dinda Aulia Putri Br Sbr — Thu, 15 Oct 2026 00:00:00 +0700

Delayed BPJS claim payments can disrupt hospital financial management, requiring prediction based on historical data. This study applies an Artificial Neural Network using the Backpropagation algorithm to predict delayed BPJS claim payment status in Binjai City. Claim data from 2023–2024 were used for training, while 2025 data were used for testing. The model employed a 4–3–1 architecture, a learning rate of 0.2, a maximum of 10,000 epochs, and a target error of 0.01 using MATLAB. Training produced a Mean Squared Error of 0.0091831 at the 87th epoch. Testing on 12 records yielded 11 correct predictions and one incorrect prediction, achieving 91.67% accuracy. These results indicate Backpropagation performs well and may support faster, objective, consistent, measurable claim payment evaluation in hospitals.

A Comparative Study of Naive Bayes and Decision Tree Classifiers for Forecasting Attrition Risk Among AI Professionals Based on Burnout Indicators

Aprian Java Dillah, Budi Santoso — Thu, 15 Oct 2026 00:00:00 +0700

Employee turnover in artificial intelligence (AI)-related roles is closely linked to burnout, yet many organizations still lack a systematic way to flag at-risk staff before they resign. This study addresses that gap by building and comparing two supervised classifiers, Naive Bayes and Decision Tree, to forecast attrition risk from workplace burnout signals. A publicly available Kaggle dataset comprising 1,500 employee records across 19 attributes was cleaned through mean/mode imputation and narrowed down via feature selection before being split into training and testing subsets. Using Orange Data Mining together with 10-fold cross-validation, the two models were benchmarked twice: once on the full feature set and once after irrelevant attributes were removed. Prior to feature selection, Decision Tree slightly led with 78.8% accuracy against Naive Bayes' 77.6%; after selection, the ranking reversed, with Naive Bayes reaching 80.2% accuracy versus 80.1% for Decision Tree. The comparison suggests that neither algorithm is universally superior performance instead hinges on how the underlying features are curated, a finding that should inform how HR analytics teams design early-warning systems for burnout-driven attrition.

A User Fatigue Prediction System Based on Heart Rate and SpO₂ Data from Smartwatch on Android Platform

Ridho Nur Fahrendy fahrendy, Fara Triadi, Agus Triyono — Thu, 15 Oct 2026 00:00:00 +0700

The high demands of modern activities frequently lead to fatigue, a condition that diminishes physical and cognitive capacity while elevating the risk of human error. Smartwatches equipped with physiological sensors offer continuous, non-invasive health monitoring, presenting an opportunity for early fatigue detection. This study developed an Android application that predicts user fatigue status by analyzing heart rate (BPM) and blood oxygen saturation (SpO₂) data recorded by a smartwatch. A machine learning model based on Long Short-Term Memory (LSTM) was trained on a public dataset to recognize fatigue patterns, then converted to TensorFlow Lite format for on-device integration. The application retrieves historical BPM and SpO₂ data through Health Connect API and classifies the user's condition as either Normal or Fatigued. Offline evaluation yielded 100% accuracy on test data; however, this performance was identified as an artifact of overfitting caused by deterministic synthetic labeling rules. On-device inference encountered compatibility issues with TensorFlow Lite runtime, preventing successful execution of the LSTM model on Android. Despite this limitation, the study demonstrates the feasibility of bridging heterogeneous wearable ecosystems and highlights critical technical challenges in deploying recurrent neural networks on mobile devices.

Prototype of a mini wind turbine for automatic charging based on the Internet of Things (IoT) in the garden area of STMIK Kaputama

Deva Ryandi, Achmad Fauzi, Husnul Khair — Thu, 15 Oct 2026 00:00:00 +0700

Lighting in the garden area of STMIK Kaputama Binjai, located on the upper floor of the building, remains very limited and fully dependent on the conventional power supply, making the area uncomfortable and potentially unsafe for use at night. This research aims to design and build a mini wind turbine prototype as an alternative energy source for garden lamps, to develop an automatic charging system that stores wind energy in a battery and switches the lamp on based on a light sensor, and to implement Internet of Things (IoT) technology through the Blynk application for remote monitoring and control. The research was developed using the prototype method, with a system consisting of a vertical axis wind turbine (VAWT) coupled to a 12 V DC generator, a 1N5822 Schottky diode, an LM2596 module, a TP4056 module as the charging controller for an 18650 battery, an XL6009 module as a step-up converter for the LED lamp, a voltage sensor, an INA219 current sensor, an LDR sensor, a relay, and a NodeMCU ESP32 as the main controller. Testing was conducted directly in the garden area of STMIK Kaputama using naturally available wind. The results show that the prototype produced 0.11-2.62 W of power at wind speeds of 1.8-5.3 m/s, the automatic charging system raised the battery voltage by 0.48 V (from 3.62 V to 4.10 V) over three hours, the LDR-based automatic lighting control succeeded in all five lighting conditions (100%), the charged battery powered the lamp for more than six hours, and real-time monitoring via Blynk operated with an average delay of 1.40 seconds with remote ON/OFF control succeeding in 10 out of 10 attempts, supported by an average error of 1.74% for the voltage sensor and 3.06% for the INA219 current sensor. Therefore, the IoT-based mini wind turbine prototype is proven capable of serving as an independent and environmentally friendly alternative energy source for campus garden lighting, although the charging rate is still affected by fluctuations in local wind speed.

Analysis of Hoax News in News Media Using the K-Nearest Neighbor (K -NN) Method

Putri Nabila , Bustami, Wahyu Fuadi — Thu, 15 Oct 2026 00:00:00 +0700

The rapid development of internet technology has facilitated the dissemination of information within the community through various news media. However, this convenience has also fueled the widespread dissemination of unverified fake news (hoaxes), potentially causing social unrest, undermining the credibility of information, and manipulating public opinion. Therefore, a computing system capable of automatically filtering and categorizing news is needed to help readers distinguish between facts and hoaxes. This study aims to develop an automatic classification model to identify hoax news using the K-Nearest Neighbor (KNN) method and measure its accuracy.

This study uses the Euclidean Distance approach with a value of k = 3. This study uses 2,000 news data that have been collected and verified through official fact-checking sites. The data was divided using the train-test split method with a ratio of 80:20 to obtain 1,600 training data and 400 testing data . Based on the results of the confusion matrix test at a value of K = 3, the True Positive (TP) value was 104, False Positive (FP) was 0, False Negative (FN) was 95, and True Negative (TN) was 201. The model performance evaluation showed a precision value of 100%, recall of 52.26%, and accuracy of 76.25%. These results indicate that the K-NN method is able to classify news well and has a high level of precision, although the model's ability to detect all hoax news ( recall ) still needs to be improved in further research. This research is expected to contribute as a technology-based tool ( decision support system ) for the public and academics in monitoring, filtering, and tackling the spread of fake news in digital media.

Design and Implementation of a Web-Based Cadre Data Management Information System for HMI Tembilahan Branch

Moh Idris, Abdul Muni, Samsudin — Thu, 15 Oct 2026 00:00:00 +0700

This study designs and implements a web-based information system for managing cadre data at the Himpunan Mahasiswa Islam (HMI) Tembilahan Branch. The study responds to handwritten and fragmented administrative practices that slow data retrieval, increase duplication risk, and weaken information control. Data were collected through direct observation at the branch secretariat, an interview with the branch chairman, and a literature review. System requirements were analyzed using the PIECES framework, while development followed the Waterfall model and was represented through Unified Modeling Language diagrams. The application was implemented with Laravel and MySQL and provides role-based access for administrators and the HMI chairman, covering member, commissariat, announcement, activation, position, and profile management. Functionality testing involved 25 respondents and 25 system functions, producing 625 successful executions and no failures, equivalent to a functionality value of 1. Usability testing with ten statements produced 1,095 of 1,250 possible points, or 87.64%, categorized as very good. The results indicate that the system is functionally reliable, usable, and suitable for supporting more organized cadre administration at HMI Tembilahan Branch.

Internet of Things System Implementation for Post-Landfill Soil Rehabilitation Based on pH and Moisture Monitoring

Muhammad Haikal Al Qalbi , Wahyuni Eka Sari , Abdullah Hanif — Thu, 15 Oct 2026 00:00:00 +0700

Post-landfill land is characterized by extreme soil conditions, such as low acidity (pH) and poor water retention, which impede rehabilitation and revegetation processes. Conventional manual watering and fertilization methods have proven inefficient in responding to these dynamic and degraded soil conditions. This study aims to design and implement an Internet of Things (IoT) system to support land rehabilitation through automated monitoring and control of irrigation and fertilization (pH adjustment). The system is built using an ESP32 microcontroller integrated with a capacitive soil moisture sensor, a PH-4502C pH sensor, a TDS sensor, and a DHT11 sensor to monitor critical soil and environmental parameters in real time. Based on the sensor data, the system autonomously activates a water pump and a fertilizer pump according to predefined threshold logic, and all telemetry data and actuator statuses are transmitted to a Firebase-based interactive web dashboard that displays numerical readings, trigger-logic status, and historical trend graphs. Testing was carried out through sensor-accuracy validation, control-logic functionality testing, and interface testing. The results show that the soil-moisture sensor achieved an average relative error below 8% and the pH sensor below 3.2% compared to standard laboratory instruments, while the automatic control logic consistently activated the irrigation and fertilization pumps in accordance with the defined threshold rules. These findings indicate that the developed system can reliably monitor critical soil parameters and perform automated, data-driven interventions, offering a practical solution to improve the efficiency of water and nutrient use and to support the revegetation of post-landfill critical land.

Clustering of Community Complaints on Flooding and Waterlogging Using the K-Means Clustering Method at the Public Works and Spatial Planning Agency

Rahmawati Dewi, Magdalena Simanjuntak, I Gusti Prahmana — Thu, 15 Oct 2026 00:00:00 +0700

The Department of Public Works and Spatial Planning (PUTR) plays an important role in infrastructure development, which has a significant impact on regional economic growth. Flooding and waterlogging problems often occur due to high rainfall and suboptimal drainage systems, resulting in losses and disruptions to community activities.This study aims to classify public complaints related to flooding and waterlogging using the K-Means Clustering method. The data used consists of 617 public complaints, which are grouped into three clusters. The results show that Cluster 1 contains 355 data points, Cluster 2 contains 94 data points, and Cluster 3 contains 168 data points. This clustering is expected to assist PUTR in understanding patterns of public complaints and in determining more effective and targeted handling priorities.

IoT-Based Early Fire Detection System using Flame and Gas Sensors with Telegram Bot and Blynk Dashboard Integration

Novi Ramadani — Thu, 15 Oct 2026 00:00:00 +0700

Fire incidents continue to pose significant risks to human safety and property due to delayed detection and response. Conventional fire alarm systems generally provide only local alerts, limiting remote monitoring and timely emergency actions. This study presents an Internet of Things (IoT)-based early fire detection system using an ESP32 microcontroller integrated with a KY-026 flame sensor and an MQ-2 gas sensor to detect fire and flammable gas conditions in real time. Sensor calibration was conducted to determine appropriate threshold values, resulting in a flame sensor threshold below 2000 ADC and a gas sensor threshold above 500 ADC. The proposed system integrates the Blynk platform for remote monitoring and Telegram Bot for automatic emergency notifications. To provide an initial fire suppression response, a relay activates a water pump connected to a sprinkler after continuous flame detection for six seconds, while a buzzer immediately alerts nearby occupants. Experimental results demonstrate that all hardware and software components operated reliably and communicated successfully through the Internet. The developed system provides accurate early fire detection, real-time monitoring, rapid warning notifications, and automatic initial fire suppression, making it suitable for residential and small-scale indoor fire safety applications.

Application of the K-Means Algorithm for Clustering Social Assistance Recipients (Case Study: Binjai City Social Agency)

Meliyani, Husnul Khair, Magdalena Simanjuntak — Thu, 15 Oct 2026 00:00:00 +0700

Social assistance distribution requires accurate data management to ensure that assistance reaches eligible beneficiaries. The Binjai City Social Agency manages the National Socio-Economic Single Data (DTSEN), which contains socio-economic information but is primarily used for administrative purposes. This study aims to apply the K-Means clustering algorithm to identify patterns among social assistance recipients based on the National Decile, Family Hope Program (PKH) recipient status, and Basic Food Assistance recipient status. The research employed the Knowledge Discovery in Databases (KDD) methodology, consisting of Data Selection, Data Preprocessing, Transformation, Data Mining, and Evaluation. A total of 847 DTSEN records were analyzed using MATLAB, while the optimal number of clusters was determined using the Elbow Method. The results indicated that the optimal number of clusters was three, representing recipient groups with different socio-economic characteristics based on National Decile and social assistance status. The clustering results provide useful information for understanding recipient characteristics and can support the Binjai City Social Agency in managing social assistance data more effectively.

Sentiment Analysis of Duolingo Application Reviews on Google Play Using the Support Vector Machine Method

Nur Jihan Arifi, Novriyenni, Muammar Khadapi — Thu, 15 Oct 2026 00:00:00 +0700

Abstract

The Duolingo app is one of the most widely used language-learning apps available on Google Play. The large number of reviews posted by users makes sentiment analysis necessary to identify trends in user opinions regarding the app—whether positive, negative, or neutral. This study aims to analyze the sentiment of Duolingo app user reviews on Google Play using the Support Vector Machine (SVM) algorithm. Research data was obtained through web scraping of 1,220 user reviews, followed by preprocessing, sentiment labeling, and word weighting using TF-IDF prior to the classification process using the SVM algorithm. The results show that positive sentiment accounted for 57.1%, negative sentiment for 30.3%, and neutral sentiment for 12.6%. Based on the test results, the Support Vector Machine (SVM) algorithm achieved an accuracy of 79.23%, demonstrating fairly good performance in classifying the sentiment of user reviews for the Duolingo app on Google Play.

Keywords: Sentiment Analysis, Duolingo, Google Play, Support Vector Machine (SVM).

Implementation of Certainty Factor Method in a Web-Based Expert System for Abacus Student Level Assessment

Hani Harmaini, Magdalena Simanjuntak, Siswan Syahputra — Thu, 15 Oct 2026 00:00:00 +0700

The placement of new students into appropriate program levels is a critical factor in the effectiveness of mental arithmetic education through the abacus method. At SEMPOA SIP Binjai Mas, this process has historically been conducted manually, relying heavily on the subjective intuition of instructors, which often results in inconsistent decision-making and assessment bias. Furthermore, the inherent uncertainty in student performance during diagnostic tests often caused by nervousness or incomplete understanding further complicates accurate classification. This research aims to design and implement a web-based expert system using the Certainty Factor (CF) method to provide a standardized, objective, and mathematically grounded recommendation for student placement. The system was developed using the Waterfall methodology and utilizes 11 validated capability indicators to categorize students into two primary levels: Junior and Foundation. The Certainty Factor method was specifically employed to quantify uncertainty and manage the degrees of confidence in student performance by integrating expert knowledge (MB and MD values) with user-provided observations. Empirical validation through a case study demonstrated high precision, yielding a 99.73% confidence level for program placement in a scenario where manual assessment might prove ambiguous. Black-box testing further confirmed that all system functionalities are 100% valid and meet operational requirements. The implementation of this expert system significantly mitigates assessment subjectivity, enhances administrative efficiency through real-time data access, and establishes a transparent, professional assessment framework for the institution.

Design and Implementation of an Internet of Things System for Air Conditioner Monitoring and Control

Wilfred Brenaldo, Agus Triyono, Fara Triadi — Thu, 15 Oct 2026 00:00:00 +0700

An Air Conditioner (AC) is widely used to keep indoor spaces comfortable, but its operation is commonly managed manually through a remote control, which often leads to operational negligence such as forgetting to switch off the unit when a room is empty. This negligence results in energy waste and increased electricity costs, a problem that is particularly relevant for spaces with high-intensity, scheduled usage such as houses of worship. This study aims to design and build an Internet of Things (IoT) system for automatic and remote monitoring and control of an AC unit, using the Kibaid Church congregation in Samarinda as a case study. The system is built around an ESP32 microcontroller integrated with a DHT22 sensor for temperature and humidity monitoring, a PZEM-004T module for real-time power-consumption monitoring, a rain-drop sensor for water-leak detection, and a Solid State Relay (SSR) for switching the AC unit on or off, with a local 16x4 LCD display and the Blynk mobile application for remote monitoring, manual control, and push notifications. The DHT22 temperature sensor was first calibrated against the AC thermostat setpoint, and functional testing was then carried out on nine core features in a controlled pre-testing environment prior to on-site deployment. The calibration results show an average temperature deviation of only 0.17 degrees Celsius, and all nine functional test scenarios—covering sensor readings, leak detection, local display, cloud data transmission, automatic and manual AC control, high-temperature notification, and a 24-hour endurance run—were completed successfully. These findings indicate that the developed system can reliably automate AC operation, provide real-time remote monitoring, and help reduce energy waste caused by operational negligence.

Application of the Apriori Algorithm for Discovering Patterns of Success Factors in Drug Rehabilitation: A Case Study at MAB Rehabilitation Center

Regitaa Ramadhanty, Magdalena Simanjuntak, Suci Ramadani — Thu, 15 Oct 2026 00:00:00 +0700

Drug abuse is a serious issue that requires effective rehabilitation programs. MAB Rehabilitation Center possesses rehabilitation data that have mainly been used for administrative purposes and have not been analyzed to identify patterns associated with rehabilitation success. This study aims to identify relationship patterns among factors influencing rehabilitation outcomes using the Apriori Algorithm, determine the main factors forming association rules based on support, confidence, and lift ratio, and provide recommendations for rehabilitation strategies. The dataset consists of client rehabilitation records from 2020 to 2025, including variables of drug type, frequency of drug use, dependency level, therapy attendance, family support, and rehabilitation success status. Data were processed using Google Colab with a minimum support threshold of 15% and a minimum confidence threshold of 60%. The analysis generated several association rules related to rehabilitation success. The strongest rule was Moderate Dependency Level → Successful Rehabilitation, with a support value of 22.38%, a confidence value of 63.70%, and a lift ratio of 1.05, indicating the strongest association among the identified patterns. These findings demonstrate that the Apriori Algorithm is effective in discovering meaningful association patterns and can support data-driven decision-making, client assessment, and the development of more targeted rehabilitation strategies at MAB Rehabilitation Center.

Application of Multiple Linear Regression to Predicted the Duration of Outpatient Rehabilitation in Drugs Users (Case Study: MAB Rehabilitation Center)

Nur Hanisa, Magdalena Simanjuntak, Kristina Annatasia Br.Sitepu — Thu, 15 Oct 2026 00:00:00 +0700

Narcotics abuse is a major issue that requires appropriate rehabilitation programs. The variation in outpatient rehabilitation duration among patients highlights the need for a prediction model to support rehabilitation planning. This study aims to apply the Multiple Linear Regression method to predict outpatient rehabilitation duration based on age, duration of narcotics use, frequency of use, and type of narcotic substance. The study used secondary data consisting of 800 outpatient records from the MAB Rehabilitation Center. Data processing was performed using Python in Google Colab with the Ordinary Least Squares (OLS) method through data preprocessing, dataset splitting into 80% training data and 20% testing data, and model evaluation using Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), the coefficient of determination (R²), the t-test, and the F-test. The results show that the Multiple Linear Regression method can be applied to develop a prediction model for outpatient rehabilitation duration. However, the t-test and F-test indicate that all independent variables have no significant effect on rehabilitation duration. The R² value of 0.005 indicates that the model explains only 0.5% of the variation in rehabilitation duration.

Public Sentiment Analysis Toward Indonesian House of Representatives Member Allowances using Naïve Bayes

Khairul Siregar — Thu, 15 Oct 2026 00:00:00 +0700

This study analyzes public sentiment toward allowances for members of the Indonesian House of Representatives using a Multinomial Naïve Bayes classifier. The initial dataset contained 14,000 public posts and comments collected from X and YouTube between July 2022 and February 2026. After validation and preprocessing, 12,869 records were retained. Sentiment labels were generated through an Indonesian lexicon-based procedure with two-token negation handling, while textual features were represented using augmented TF-IDF. The dataset was divided into 10,295 training records and 2,574 testing records. The resulting labels comprised 10,675 negative, 1,912 positive, and 282 neutral records. Testing produced an accuracy of 84.50%, macro precision of 0.5410, macro recall of 0.3767, and macro F1-score of 0.3819. The negative class achieved the strongest performance, whereas the neutral class was not successfully recognized because of severe class imbalance. The classification results were deployed through a Streamlit application that presents sentiment distribution, class-specific word clouds, temporal and source distributions, evaluation metrics, and new-text prediction. The findings show that Naïve Bayes can provide a useful overall classification result, but balanced class representation remains essential for reliable performance across all sentiment categories.

Clustering Outpatient and Inpatient Medical Records Based on Patient Complaints using K-Means

Nery Adelia Br Purba, Magdalena Simanjuntak, Marto Sihombing — Thu, 15 Oct 2026 00:00:00 +0700

Medical record archives contain patterns that can support hospital service planning, but large datasets are difficult to analyze manually. This study developed a clustering application to group outpatient and inpatient medical records at Delia Hospital based on age, patient complaint, and type of service. A total of 600 medical records from 2025 were selected, cleaned, transformed into numerical codes, and processed in MATLAB using the K-means algorithm with squared Euclidean distance. Three scenarios were evaluated, namely K = 3, K = 4, and K = 5, using cluster variance and the distribution of members as comparison measures. The cluster variance decreased from 4.7356 at K = 3 to 2.8841 at K = 4 and 2.0662 at K = 5. The five-cluster solution produced 129, 95, 122, 127, and 127 members, respectively, with dominant complaints of asthma, vertigo, diabetes, fractures, and gastritis. The results indicate that K = 5 provides the most compact and detailed grouping. The resulting information can support the evaluation of outpatient services, inpatient capacity, and hospital resource planning.

Implementation of the Naive Bayes Method for Sales Prediction at UD. Noer Berkah

Rosalia Bafiano Salma, Retno Wardhani, Masruroh — Thu, 15 Oct 2026 00:00:00 +0700

Sales prediction is essential for supporting production planning and inventory management in business operations. This study aimed to implement the Naïve Bayes classification method to predict sales status at UD. Noer Berkah using historical sales data. A total of 730 sales records collected from October 2024 to September 2025 were processed through data cleaning and preprocessing, resulting in 240 valid records. The dataset was divided into 192 training records and 48 testing records using an 80:20 hold-out method. The prediction model used month, work shift, meatball sales, and frozen food sales as input attributes, while sales status was used as the target class. Model performance was evaluated using a Confusion Matrix based on accuracy, precision, and recall. The model achieved an accuracy of 75%, with precision values of 85.71%, 66.67%, and 100%, and recall values of 70.59%, 90.91%, and 44.44% for the highly sold, sold, and unsold classes, respectively. These results indicate that the Naïve Bayes method provides satisfactory classification performance for identifying sales patterns from historical transaction data. The developed model can support production planning, inventory management, and business decision-making while reducing the risk of overstocking and stock shortages.

Clustering of Women's Family Planning Contraceptive Use Data Using the Clustering Method at the Population Control and Family Planning Office

Sintami Fadila, Magdalena Simanjuntak, Marto Sihombing — Thu, 15 Oct 2026 00:00:00 +0700

The Department of Population Control and Family Planning (DPPKB) of Binjai City is a government institution responsible for implementing the Family Planning (KB) Program to improve family welfare and control population growth. The increasing amount of contraceptive usage data each year requires an appropriate data processing method to generate information that supports decision-making. This study aims to apply the K-Means Clustering algorithm to classify contraceptive usage data among women of reproductive age based on age, contraceptive method, and region. The research method includes transforming the data into numerical form, performing the clustering process using three clusters (K = 3), and implementing the system using MATLAB. The results show that the K-Means Clustering algorithm successfully groups the data into three clusters based on the similarity of their characteristics. The clustering results provide information on contraceptive usage patterns according to age, contraceptive method, and region, which can support the Department of Population Control and Family Planning of Binjai City in evaluating, formulating policies, and planning Family Planning programs more effectively and accurately.

Clustering the Severity Levels of Hypertensive Patients Based on Blood Pressure using the K-Means Method

Lia Anggraini — Thu, 15 Oct 2026 00:00:00 +0700

Hypertension is a noncommunicable disease that is on the rise at the Anugrah Binjai Primary Care Clinic; however, the use of patient medical records has so far been limited to administrative purposes. This study aims to objectively and systematically classify the severity of hypertension based on data using data mining techniques. The algorithm applied is K-Means Clustering, utilizing the variables of gender, age, and systolic blood pressure from patients’ medical records. The clustering process was processed and tested using the MATLAB application through Euclidean distance calculations until convergence was achieved. The results of the study using a three-cluster scenario produced a structured division of patient groups, namely Cluster 1 (moderate severity) with 151 data points, Cluster 2 (high severity) with 151 data points, and Cluster 3 (low severity) with 215 data points. An evaluation of clustering quality using cluster variance yielded values of 0.9359 for 3 clusters, 0.8412 for 4 clusters, and 0.7132 for 5 clusters. These results demonstrate that the K-Means method is capable of mapping the severity patterns of hypertensive patients to assist clinics in designing targeted educational programs and medical management.

Internet of Things-Based White-Rumped Shama Egg Incubator

Arya Lanang Buana, Wahyuni Eka Sar, Abdullah Hanif — Thu, 15 Oct 2026 00:00:00 +0700

The Magpie Robin is a songbird species with high economic value and is widely cultivated by the community. However, the success rate of hatching of this bird's eggs is still low due to temperature and humidity instability during the incubation process. This study aims to design an Internet of Things (IoT)-based Magpie Robin egg incubator that is able to monitor and control temperature and humidity automatically and in real-time. This system uses an ESP32 microcontroller as a control center connected to a DHT22 sensor, an incandescent lamp as a heater, a DC fan as a cooler, a synchronous motor for automatic egg turning, and an ESP32-CAM camera for visual monitoring inside the incubator. Data is sent to the Blynk IoT platform for remote monitoring via a web dashboard. The development method used is Waterfall, including the stages of requirements analysis, design, implementation, and testing. Based on the results of this test, the tool is able to control the temperature automatically in the range of 37–38°C and humidity of 50–60%, in accordance with the hatching standards for Murai Batu bird eggs, the egg incubator machine operates for 14 days with 4 Murai Batu bird eggs, on the day of hatching 3 eggs successfully hatched and 1 did not hatch, eggs that did not hatch were indicated to be infertile or unfertilized. With this, the Murai Batu egg incubator machine has a success rate in hatching reaching 75%.

Design and Construction of an Automatic Rice Crackers (Rengginang) Dryer Based on the Internet of Things

Muhammad Aira Syauqi, Fara Triadi, Wahyuni Eka Sari — Thu, 15 Oct 2026 00:00:00 +0700

The traditional process of drying rice crackers still depends heavily on weather conditions, especially sunlight, which often becomes an obstacle during the rainy season. This research aims to design and develop an automatic rice crackers (rengginang) dryer based on the Internet of Things (IoT) that can operate autonomously and be monitored through a website. The system uses an ESP32 microcontroller as the main controller, a DS18B20 temperature sensor to detect temperature, heaters as heating elements, and fans to distribute and cool the air inside the drying chamber. Temperature control is performed automatically using a Decision Tree algorithm, allowing the system to turn the heaters and fans on or off according to the detected temperature conditions. The temperature data and system status are sent in real time to a Firebase-backed website, enabling users to remotely monitor the drying process. Based on testing results, the device can maintain a stable temperature between 55 and 80 degrees Celsius with a reliable control response, effectively reducing drying time to approximately four hours compared to the one-to-two-day duration required by the traditional manual method. This system is expected to

Implementation of Internet of Things Devices in Website-Based Smart Homes

M. Naja Annajafi Asyfahani, Fara Triadi, Agus Triyono — Thu, 15 Oct 2026 00:00:00 +0700

The rise of the Internet of Things (IoT) has accelerated the implementation of smart home systems, enabling household devices to be monitored and controlled remotely. This study focuses on developing a web-based smart home system using web technology, integrating Django, Supabase (PostgreSQL), Raspberry Pi, and Message Queuing Telemetry Transport (MQTT) protocol. The proposed system is implemented with a database-driven control approach where user commands are stored in the Supabase (PostgreSQL) cloud database and executed by Raspberry Pi via MQTT, while sensor data are synchronized and displayed on the web dashboard. The system also stores the historical activity of the devices, which is then analyzed in Google Colab to estimate the energy consumption. The results of the implementation show that the proposed system works successfully according to the designed architecture. Sensor data are monitored in real-time, actuators respond correctly to commands from users, and historical data are stored consistently for later analysis. The combination of Django, Supabase, Raspberry Pi, and MQTT shows that it is possible to have a unified, web-based smart home platform that provides real-time monitoring, remote control of devices, historical data management, and energy consumption analysis.

Analysis of Family Status at Risk of Stunting in Babat District using the K-Means Clustering Method

Susmita Amelia — Thu, 15 Oct 2026 00:00:00 +0700

Stunting is a national public health priority because it affects children's physical growth, cognitive development, and the quality of future human resources. To support stunting prevention efforts, it is important to identify areas with similar risk characteristics so that intervention programs can be implemented more effectively. This study aims to analyze the clustering of villages in Babat Regency based on stunting risk using the K-Means clustering algorithm. The variables used include the number of infants under two years of age, the number of children under five years of age, the number of pregnant women, and environmental factors represented by access to proper sanitation and clean water sources. A total of 21 villages were analyzed and grouped into three clusters representing high, medium, and low stunting risk. The clustering results showed that five villages were classified as high risk, four villages as medium risk, and 12 villages as low risk. These results were then compared with the original village data and showed that the clustering accurately represented the actual conditions, indicating that the K-Means algorithm effectively identified villages with similar stunting risk characteristics. These findings indicate that K-Means clustering can provide valuable information for local governments in determining intervention priorities, allocating resources efficiently, and improving the effectiveness of stunting prevention and management programs in Babat Regency.

Comparison of Naïve Bayes and Random Forest Algorithms for Customer Churn Prediction in Telecommunication Services

Hikmal Maulana, Achmad Nurhadi, Aji Sumandito — Thu, 15 Oct 2026 00:00:00 +0700

Customer churn refers to the phenomenon of customers discontinuing a service, posing a serious challenge for telecommunication companies because it directly affects revenue and company reputation. This study compares the performance of the Naïve Bayes and Random Forest algorithms in predicting customer churn in telecommunication services and analyzes the effect of feature selection on their performance. The dataset used is the Telco Customer Churn dataset obtained from Kaggle, consisting of 7,043 customer records with 21 attributes. A supervised learning approach was employed, with preprocessing stages including data cleaning, feature selection using the Gain Ratio method, and data splitting into 80% training and 20% testing data. Modeling and evaluation were conducted using Orange Data Mining software. The results show that Random Forest outperformed Naïve Bayes after feature selection, achieving an accuracy of 0.783, precision of 0.773, recall of 0.783, and F1-score of 0.776, while Naïve Bayes achieved an accuracy of 0.747, precision of 0.801, recall of 0.747, and F1-score of 0.760. Feature selection improved Naïve Bayes performance while slightly reducing Random Forest performance. The Contract, OnlineSecurity, and TechSupport attributes were identified as the most influential features for predicting customer churn.