METHODIKA: Jurnal Teknik Informatika dan Sistem Informasi

RANCANG BANGUN SISTEM LEARNING HUB BERBASIS WEB UNTUK MENDUKUNG PROSES PEMBEKALAN PESERTA MAGANG DI DIGIMIZU DIGITAL MANAGEMENT

Raihan Canggih Panilih, Marta Ardiyanto, Mira Erlinawati — Thu, 10 Sep 2026 00:00:00 +0700

The onboarding phase at Digimizu Digital Management plays a vital role in helping interns acclimate to the professional digital work environment. Currently, however, this phase encounters notable operational hurdles because it heavily depends on manual coordination via WhatsApp. This manual approach causes inconsistent material distribution, unorganized portfolio submissions, and makes it challenging for mentors to track intern development accurately. To address these issues, this study focuses on designing and implementing a centralized, web-based Learning Hub platform. The software development life cycle employed is the Waterfall model, powered by the Laravel framework and a MySQL database architecture. The platform incorporates several core functionalities: a sequential Learning Path to ensure structured knowledge absorption, a DG-XP Gamification and Leaderboard system to boost user engagement, and a Digital Portfolio module for objective performance assessment. Findings reveal that this Learning Hub effectively modernizes the onboarding workflow, reduces repetitive tasks for mentors, and establishes a clear, quantifiable record of intern achievements. Additionally, functionality validation through Black-Box Equivalence Partitioning across 10 fundamental test cases achieved a 100% success rate. This confirms that all essential mechanisms such as access gatekeepers, gamification point calculations, and automatic E-Certificate generationfunction flawlessly according to the initial system requirements.

SISTEM PREDIKSI LONJAKAN HARGA PANGAN BERBASIS RANDOM FOREST UNTUK EARLY WARNING SYSTEM

Reihan Setya Banda Syah Putra, Sopingi, Aprilisa Arum Sari — Thu, 10 Sep 2026 00:00:00 +0700

Food prices are essential indicators of economic stability and public welfare. Uncontrolled fluctuations, particularly sudden spikes, lower purchasing power and drive inflation. Currently, food price information systems are largely descriptive and lack predictive capabilities for sudden anomalies. To address this gap, this research develops a web-based Early Warning System (EWS) featuring digital integration and automation to proactively detect price spikes. The study utilizes daily secondary time-series data from the Strategic Food Price Information Center (PIHPS) via Kaggle, spanning from 2022 to 2026. Feature engineering, including 7-day moving averages and percentage changes, was applied to enhance the Random Forest classification algorithm. To handle the extreme 95:5 data imbalance, a class-weight balancing technique was employed during modeling. The empirical findings demonstrate that the model achieved an overall accuracy of 74% and a recall of 56% for the minority 'spike' class, proving its capability to capture more than half of the actual market crises. Furthermore, feature importance analysis revealed that the 7-day moving average was the most dominant predictor, contributing 26.22% to the model's decisions, which indicates the system effectively recognizes historical market volatility rather than nominal values. This prediction engine was successfully integrated into an interactive Laravel-based dashboard equipped with automated alert notifications. However, the system's high sensitivity resulted in a low precision of 9%, generating frequent false positive alerts that require further architectural refinement to mitigate alert fatigue in future studies.

ANALISIS SENTIMEN PUBLIK TERHADAP KABINET MERAH PUTIH PADA APLIKASI X DENGAN METODE BIDIRECTIONAL ENCODER REPRESENTATION FROM TRANSFORMER

Darwis Robinson Manalu, Jusup Sihotang, Edward Rajagukguk — Thu, 10 Sep 2026 00:00:00 +0700

The X application has become a platform for Indonesians to express their opinions, including those related to the Kabinet Merah Putih. This study aims to analyze public sentiment on the X application regarding the Kabinet Merah Putih using the IndoBERT method. A total of 1944 tweets were collected from the X platform using the keyword “Kabinet Merah Putih”, then processed through preprocessing stages including cleaning, normalization, case folding, and stopword removal, automatic labeling with TextBlob, and data division for model training and testing. The finetuned IndoBERT model was used to classify sentiment into positive, negative, and neutral. The results of the study show that public sentiment is dominated by positive and neutral sentiments, while IndoBERT's performance shows good accuracy in sentiment classification, with a model accuracy value of 85.96%, precision of 75%, recall of 74%, and F1-Score of 74%. This study provides an objective picture of public perception of the Kabinet Merah Putih on the X application and is expected to serve as evaluation material for the government.

SISTEM INFORMASI MANAJEMEN ASET DAN PERSEWAAN SOUND SYSTEM BERBASIS WEB MENGGUNAKAN METODE FIFO UNTUK OPTIMALISASI PEMANFAATAN ASET

Abim Febri Hananto, Aprilisa Arum Sari, Agustina Purwatiningsih — Thu, 10 Sep 2026 00:00:00 +0700

KF Sound System faces operational challenges due to a manual asset management system, which leads to uneven equipment wear and tear and difficulties in monitoring inventory availability. This study aims to design a web-based Asset Management Information System that integrates customer orders, inventory management, and payment automation to enhance operational effectiveness. The system development applies the prototyping method using the PHP language and the Laravel framework. The main innovation in this study is the implementation of the First-In, First-Out (FIFO) algorithm for dynamic asset allocation, where the system automatically prioritizes equipment with the longest idle time to ensure that asset depreciation is distributed evenly. Test results indicate that this integrated platform successfully streamlines workflows, prevents 100% of double-booking cases, and accelerates the equipment allocation processing time by up to 80% compared to manual methods. However, this system still has limitations, namely the absence of automatic hardware damage detection (IoT), meaning that updating the physical eligibility status of equipment post-rental remains entirely dependent on manual input by the administrator.

PENGEMBANGAN SISTEM MONITORING DAN REKAPITULASI PROYEK BERBASIS WEB UNTUK OPTIMASI WORKFLOW DI DIGIMIZU DIGITAL MANAGEMENT

DWI WAHYUNINGTYAS, HANIFAH PERMATASARI, AGUSTINA SRIRAHAYU — Thu, 10 Sep 2026 00:00:00 +0700

This study aims to develop a web-based project monitoring and recapitulation system to support the monitoring and reporting processes at Digimizu Digital Management. The main problem addressed is that project management is still carried out manually through unstructured communication, making it difficult to track data and control work progress. The method used in this research is the Waterfall development model, which consists of requirements analysis, system design, implementation, and testing stages. The system was developed using the PHP programming language with the Laravel framework and MySQL as the database. System testing was conducted using the black-box testing method to ensure that all system functions operate as expected. The results indicate that the developed system is capable of integrating project management, task distribution, and progress monitoring into a centralized platform. In addition, features such as multiuser management, hierarchical validation, monitoring dashboards, and automated reporting contribute to improved efficiency, transparency, and data accuracy. Based on the testing results, all main system features function properly and meet user requirements. Therefore, this system can serve as an effective solution to enhance project management performance at Digimizu Digital Management.

PENGEMBANGAN SISTEM PEMESANAN KUE KERING BERBASIS WEB PADA UMKM DAPOER TABI

Thu, 10 Sep 2026 00:00:00 +0700

The current development of digital technology encourages Micro, Small, and Medium Enterprises (MSMEs) to transform their entire business processes, including Dapoer Tabi, a culinary pastry business that still records orders conventionally via WhatsApp and logbooks, making it prone to data recapitulation errors during high order volumes, complicating real-time stock monitoring, and lacking integrated sales reporting. This study aims to design and develop a web-based pastry ordering system to digitalize the business processes at MSME Dapoer Tabi through the implementation of the Prototyping methodology, which includes the stages of requirements gathering, quick design, prototype development, prototype testing, and refinement. Based on the testing results using the black box testing method on 9 main test scenarios, all features on the platform functioned properly without any issues and achieved a 100% functional system success rate. The implementation of this online ordering system provides significant practical benefits for Dapoer Tabi in simplifying the customer ordering process, making administrative transaction recording more efficient, and digitally expanding its marketing reach.

SISTEM PREDIKSI VOLUME PENUMPANG HARIAN KRL YOGYAKARTA-SOLO MENGGUNAKAN MODEL HYBRID SARIMAX-PROPHET

Muhammad Ilham 'Aziiz Alfarobi, Nurmalitasari, Ratna Puspita Indah — Thu, 10 Sep 2026 00:00:00 +0700

The operation of the Yogyakarta-Solo Commuter Line (KRL) since 2021 has become the backbone of transportation in the Yogyakarta Special Region and Central Java. However, highly dynamic fluctuations in passenger volume pose challenges for operational optimization. This research aims to develop an accurate daily passenger volume prediction system with a 30 day forecasting horizon to mitigate overcrowding and fleet inefficiency. The methodology employed is CRISP-DM, proposing a layered hybrid architecture based on residual modeling. In this model, SARIMAX serves as the primary pattern modeler (Layer 1), while Facebook Prophet acts as a residual corrector (Layer 2), optimized with selective correction mechanisms and daily adaptive weights. The research data covers the period from January 2025 to January 2026, totaling 396 observations. The test results show that the hybrid model provides the best performance compared to single models, achieving a Mean Absolute Percentage Error (MAPE) of 9.66% and a Mean Absolute Error (MAE) of 2,739 passengers per day. Utilizing historical data from January 2025 to January 2026, the modeling results are integrated into an interactive Streamlit dashboard as a practical decision support tool for KAI Commuter's proactive operational planning

PERANCANGAN PENJUALAN ONLINE AQIQAH DAN QURBAN BERBASIS WEB MENGGUNAKAN METODE DESIGN THINKING

Muhammad Alvarel Diyandra Alfaya Varel, Zulhalim, Balthasar Lumbantobing — Thu, 10 Sep 2026 00:00:00 +0700

Advances in information technology have encouraged businesses to utilize digital media to support marketing and sales activities, including those in the aqiqah and qurban sector. This study aims to design a web-based online sales system for Kambing Bang Muksin, located in East Jakarta, to improve promotional effectiveness and facilitate information delivery to customers. The research employed the Design Thinking method, which consists of five stages: Empathize, Define, Ideate, Prototype, and Test. Data were collected through observation, interviews with the business owner and five customers, literature review, and documentation to identify user requirements and problems encountered in the existing sales process. A website prototype was developed using HTML and CSS, featuring product catalogs, service information, and customer reviews. The prototype testing stage involved one business owner and five customers to evaluate usability, interface appearance, and feature suitability according to user needs. The test results showed that five out of six respondents (83.3%) considered the system very easy to use, while one respondent (16.7%) considered it easy to use. Furthermore, all respondents (100%) stated that the product information and available features met their needs. The results indicate that the developed website provides clearer product information, facilitates data management, expands promotional reach, and supports the digitalization process of Kambing Bang Muksin's business more effectively.

EVALUASI TATA KELOLA SISTEM INFORMASI KEUANGAN UNIVERSITAS ADVENT INDONESIA MENGGUNAKAN FRAMEWORK COBIT 2019 (DOMAIN MEA01)

Grace Aritonang, Raymond — Thu, 10 Sep 2026 00:00:00 +0700

The Financial Information System (FIS) plays an important role in managing financial data and supporting decision-making within organizations. This study aims to evaluate the governance of FIS at Universitas Advent Indonesia (UNAI). The assessment was conducted using the COBIT 2019 framework in the MEA01 area (Managed Performance and Conformance Monitoring). Questionnaires were distributed to 8 staff members working in UNAI’s finance department. The collected data were analyzed by calculating the capability level and performing a gap analysis. The results show that the governance of UNAI’s FIS is at level 3 (Established). This indicates that performance monitoring and evaluation processes are carried out regularly and are well-documented. Two practices, namely MEA01.03 and MEA01.04, reached the Fully Achieved category. Three other practices, MEA01.01, MEA01.02, and MEA01.05, were categorized as Largely Achieved. However, shortcomings remain in setting performance targets and implementing corrective actions. Recommendations for improvement include strengthening the monitoring system, regularly updating performance targets, and enhancing corrective measures to ensure better FIS governance in the future.

ANALISIS POLA PEMBELIAN PRODUK MENGGUNAKAN ALGORITMA APRIORI PADA DATA TRANSAKSI RETAIL

Romadona — Thu, 10 Sep 2026 00:00:00 +0700

The growth of transaction data in the retail sector increases the need for analytical methods capable of identifying consumer purchasing patterns efficiently. This study applies the Apriori algorithm within the Cross Industry Standard Process for Data Mining (CRISP-DM) framework to perform market basket analysis on retail transaction data. The dataset consists of 934,974 transaction records, including 407,171 unique transactions and 27,344 unique products collected between July 2021 and September 2025. After the data cleaning process, 189,724 valid transactions were obtained. To improve computational efficiency, the analysis was limited to the 300 best-selling products, resulting in 90,718 transactions for the modeling stage. Frequent itemset generation was performed using a minimum support value of 0.1% and a maximum itemset length of three, producing 570 frequent itemsets consisting of 300 1-itemsets, 213 2-itemsets, and 57 3-itemsets. Association rule generation using a minimum confidence threshold of 70% and a lift ratio greater than 1 produced 52 valid rules. The best rule achieved a lift ratio of 161.32 and a confidence value of 93.88%, indicating a strong purchasing relationship among school supply products. The results demonstrate that the selected support and confidence parameters are effective in identifying meaningful purchasing patterns. Furthermore, the resulting association rules can support practical retail strategies, including product bundling, shelf arrangement optimization, and inventory management.

PENERAPAN METODE HYBRID RANDOM FOREST DAN GENETIC ALGORITHM UNTUK OPTIMASI PENJADWALAN PRODUKSI

Widya Monika Sari, Nurmalitasari, Bangun Prajadi Cipto Utomo — Thu, 10 Sep 2026 00:00:00 +0700

The garment industry faces complex production scheduling challenges due to high product variability and inaccurate process time estimation. Inefficient production scheduling often leads to production delays and reduced operational performance. Conventional scheduling methods such as First Come First Serve (FCFS) and Earliest Due Date (EDD) have been shown to be less effective in dynamic production environments. This study aims to optimize flow shop production scheduling in a children's garment manufacturing environment using a hybrid Random Forest–Genetic Algorithm approach. Random Forest is employed to predict the processing time of each job based on a simulated dataset regenerated from the company's historical production data collected in 2025 while preserving the statistical characteristics and relationships among variables. Subsequently, the Genetic Algorithm is used to optimize job sequencing by simultaneously minimizing makespan and weighted tardiness. The study follows the CRISP-DM methodology up to the model evaluation stage. The results show that the Random Forest model achieved satisfactory prediction performance for the cutting, sewing, and finishing stages, with R² values of 0.817, 0.981, and 0.867, respectively. Using a population size of 30, 100 generations, and 10 independent runs, the Genetic Algorithm achieved an improvement of 79.94% compared to FCFS and 39.71% compared to EDD. These findings demonstrate that the proposed hybrid Random Forest–Genetic Algorithm approach can generate a more adaptive, efficient, and data-driven production schedule than conventional scheduling methods.

SISTEM INFORMASI PENYEWAAN PERALATAN OUTDOOR BERBASIS WEB DI SELEKTA ADVENTURE

Fajar Saputra, Wijiyanto, Hanifah Permatasari — Thu, 10 Sep 2026 00:00:00 +0700

The operational management of outdoor equipment rentals at Selekta Adventure still relies on manual processes, potentially leading to errors in transaction recording, delays in reporting, and difficulties in monitoring stock availability. This situation indicates the need for a system capable of effectively integrating all rental activities. This study aims to develop a web-based outdoor equipment rental information system by applying the Waterfall method as a software development approach. The system implementation was carried out using the Laravel framework and MySQL database to support structured data management. The resulting system provides various key features, including real-time inventory management, automatic late payment penalty calculations, and digital payment integration through a payment gateway, making the transaction process more practical and transparent. Functional testing was conducted using the Black Box Testing method, and all test scenarios showed that each feature ran according to user requirements without any functional errors. In addition, the application quality was evaluated using Google Lighthouse with a Performance score of 92, Accessibility 95, Best Practices 96, and SEO 100. The evaluation results indicate that the system not only meets functional requirements but also has a good web interface quality in terms of performance, accessibility, implementation of best practices, and search engine optimization. Thus, the developed system can support increased operational efficiency, reduce potential errors in data management, and improve the quality of rental services at Selekta Adventure.

PEMODELAN BANJIR DI PASTEUR JAWA BARAT MENGGUNAKAN SAINT VENANT EQUATION DENGAN METODE BEDA HINGGA

Febriana Eka Adkhaniyah, Nimas Nabila Anggraeni, Dian Candra Rini Novitasari — Thu, 10 Sep 2026 00:00:00 +0700

Floods are natural disasters that often occur in Indonesia. Floods occur due to excessive water flow that inundates an area for a certain period of time. Floods can occur suddenly or gradually at any time, resulting in significant losses. According to the National Disaster Management Agency, from 2022 to 2024, Indonesia experienced 4,026 flood disasters, 699of which occurred in West Java Province. On January 25, 2025, one of the areas in West Java, specifically Pasteur, Bandung, experienced a flood disaster triggered by the suboptimal performance of its drainage system and high rainfall intensity. Therefore, this research conducted flood modeling using the shallow water equation with the finite difference method. The purpose of this research is to minimize losses caused by flood disasters. Based on the research results, the Pasteur area has the potential to experience flooding if it experiences high rainfall intensity, which is also exacerbated by the inadequate drainage system. The modeling results indicate that the depth of the floodwater increased significantly, reaching a maximum depth of approximately 1.3 meters, after which it gradually decreased until it receded in about six hours.

KLASIFIKASI PENYAKIT JAMUR PADA TANAMAN BAWANG MERAH MENGGUNAKAN METODE CONVOLUTIONAL NEURAL NETWORK (CNN) BERBASIS CITRA DIGITAL

Alfani Septiani Selan, Franky Franky Y. Basilin, Tika Skolastika S. Igon — Thu, 10 Sep 2026 00:00:00 +0700

Shallots are a type of bulb plant widely consumed by Indonesians, both as a cooking spice and herbal medicine. Shallot production in Kupang City has experienced a significant decline, with production dropping from 292.15 quantiles to 255.01 quantiles in 2024. This is a serious concern due to disease attacks on shallot plants that cause economic losses due to crop failure for farmers. Lack of understanding and knowledge about shallot diseases is a major obstacle in overcoming this problem. Therefore, an automated system based on digital image technology is needed that is capable of classifying diseases quickly and accurately. This study aims to implement a Convolutional Neural Network (CNN) in the process of classifying fungal diseases in shallot plants based on digital images. CNN is a deep learning method that has the ability to extract visual features through convolutional, pooling, and fully connected layers. The use of CNN in this study is expected to provide accurate results in classifying fungal diseases in shallot plants based on digital images, thereby reducing the potential for crop failure and increasing production yields. The test results using K-Fold Cross Validation showed that the best model was obtained in fold 5 with an accuracy of 90.00%, a sensitivity of 90.02%, and a specificity of 96.75%. In addition, the system obtained an average accuracy of 86.91%, a sensitivity of 87.61%, and a specificity of 95.61%. Based on these results, the system is able to classify fungal diseases in shallot plants with good performance.

ANALISIS ROBUSTNESS CONVOLUTIONAL NEURAL NETWORK TERHADAP VARIASI PENCAHAYAAN PADA SISTEM PENGENALAN WAJAH

Ezra Ananta Pandie, Franki Yusuf Bisilisin, Dewi Anggraini — Thu, 10 Sep 2026 00:00:00 +0700

This study aims to evaluate the robustness of Convolutional Neural Networks (CNN) in face recognition systems under varying illumination conditions. The evaluation was conducted using a dataset comprising 36 subjects, with facial images captured under three distinct lighting scenarios: dim, normal, and bright. The research methodology involved training the CNN model using K-Fold Cross-Validation and assessing its stability against visual disturbances using artificial adversarial attacks based on the Fast Gradient Sign Method (FGSM). The novelty and main contribution of this study lie in the dual-evaluation approach, which simultaneously tests the model's resilience against natural illumination variations and artificial adversarial perturbations. Experimental results demonstrated that the CNN model achieved optimal face recognition performance at 50 epochs, maintaining an average accuracy rate of 81.48%. In conclusion, the evaluated CNN architecture is reliable and stable for face recognition in uncontrolled lighting environments, providing a solid foundation for developing more secure biometric systems against visual disturbances.

RANCANG BANGUN WEBSITE SISTEM INFORMASI INVENTARIS BARANG MENGGUNAKAN METODE WATERFALL DI DESA PADA - LEMBATA

Kornelis Andrian Kabo, Yohanis Malelak, Petrus Katemba — Thu, 10 Sep 2026 00:00:00 +0700

Inventory management at the Pada Village Office in Lembata is currently still conducted manually using physical ledgers, which triggers data inaccuracies and the risk of document loss. This study aims to analyze, design, and develop an integrated web-based inventory information system using the Waterfall method. The system implements a real-time public complaint feature to strengthen asset management transparency and support the implementation of the Electronic-Based Government System (SPBE), enriched with a public infrastructure loan/rental module and an automatic fine control mechanism. The system was built using Native PHP and a MySQL database. Functional evaluation using Black Box Testing on 8 core test scenarios showed a 100% success rate (Pass) without any technical errors. Performance efficiency evaluation through Google Chrome Network Tools demonstrated optimal server response speed, with an overall average Page Load Time of 751 ms (under 1 second). Compatibility testing verified that the public interface is fully responsive and adaptive when accessed through both desktop and mobile browsers. The integration of these various features establishes the proposed system as a practical solution for realizing accountable village-level administrative digitalization.

SISTEM PENDUKUNG KEPUTUSAN PEMBERIAN BONUS KARYAWAN MENGGUNAKAN METODE SIMPLE ADDITIVE WEIGHTING

Fahmi Prima Yasa, Ifan Junaedi, Anton Zulkarnain Sianipar — Thu, 10 Sep 2026 00:00:00 +0700

Employee bonus allocation is one form of corporate appreciation for employee performance in supporting organizational goals. However, the bonus determination process, which is still carried out manually at a private company that is the object of this study, often leads to subjectivity and inaccuracy in decision-making. This study aims to design and build a Decision Support System (DSS) for employee bonus allocation using the Simple Additive Weighting (SAW) method. The SAW method is used for multi-criteria calculation through the weighting and ranking of alternatives. The system was developed as a web-based application using the PHP programming language and MySQL database, applying six assessment criteria: attendance, discipline, work quality/competence, teamwork, loyalty, and administrative violations. The research was conducted using the System Development Life Cycle (SDLC) waterfall model, consisting of requirement analysis, system design, implementation, and testing. The system was tested on 20 employee alternative data and was able to perform objective employee assessment and ranking based on the highest preference value, which was then mapped into four bonus-recommendation zone classifications. Black Box Testing on authentication security, data integrity, computational accuracy, and interface usability showed a 100% success rate with a 0% computational error margin, indicating that all system features functioned according to user requirements. With this system, the employee bonus determination process becomes more effective, transparent, and efficient.

ANALISIS PERBANDINGAN SISTEM NOTIFIKASI TELEGRAM DAN BERBASIS WEB UNTUK PEMANTAUAN TERNAK IOT

Rizki Fikriansyah, Fera Febrianti — Thu, 10 Sep 2026 00:00:00 +0700

Although livestock monitoring is essential for improving farm productivity and supporting food security, conventional monitoring methods remain inefficient and may increase the risk of livestock loss. This study aims to compare the performance of Telegram-based and web-based notification systems for Internet of Things (IoT)-based livestock monitoring and to evaluate the responsiveness of the proposed web-based system. The developed system uses an ESP32 and Neo-6M GPS module to transmit livestock coordinates to a PHP-based web server, where data are stored in a MySQL database and displayed in real time using AJAX and a geofencing mechanism. System performance was evaluated through 12 field experiments conducted during morning, afternoon, and evening conditions. The results show that the proposed system achieved a notification delay of 0.06–3.4 s under normal network conditions, with a maximum delay of 25 s when the device temporarily moved outside the WiFi coverage area. Compared with the previous Telegram-based system, which relied on periodic notifications and external messaging services, the proposed web-based system provides faster and more responsive real-time monitoring. These findings demonstrate that the proposed approach can improve livestock supervision efficiency and has practical potential for supporting smart farming applications and food security.

SISTEM PREDIKSI RISIKO KETERLAMBATAN DISTRIBUSI PANGAN PROGRAM MAKANAN BERGIZI GRATIS BERBASIS MACHINE LEARNING

Ismail, Nur Fadillah Amiruddin, Hasna, Zinta, Kamis Tati — Thu, 10 Sep 2026 00:00:00 +0700

Food distribution in the Free Lunch Program (MBG) requires timeliness to maintain food quality and service effectiveness to beneficiaries. Delays can be influenced by distribution distance, number of recipients, weather, road conditions, delivery time, and vehicle type. This study aims to develop a machine learning-based prediction model for the risk of delays in MBG food distribution. The research method uses a quantitative approach with a dataset of 100 data samples from operational distribution scenarios in Soppeng Regency. Data were processed through cleaning, categorical variable coding, numeric variable normalization, 5-fold cross-validation splitting, model training, and performance evaluation. Four algorithms were compared: Random Forest, Decision Tree, K-Nearest Neighbor, and Logistic Regression. The test results showed that Random Forest achieved 92.00% accuracy, 92.00% precision, 92.00% recall, and 92.00% F1-score. Feature importance analysis showed that the number of recipients, distribution distance, and distribution time were the most dominant factors in determining the risk of delays. The proposed prediction system can be a tool for MBG distribution managers in identifying potential delays early and formulating more appropriate operational mitigation recommendations.

ANALISIS PENJUALAN PADA HAPPYMART MENGGUNAKAN ALGORITMA FP-GROWTH

Hendrikus Lambertho Laba Kumanireng, Franki Yusuf Bisilisin, Dewi Anggraini — Thu, 10 Sep 2026 00:00:00 +0700

Sales analysis is a crucial process for evaluating transaction data to understand consumption patterns and maximize business performance through a data-driven approach. HappyMart faces the challenge of significant transaction data growth, collecting a total of 1,500 transaction records during the period of August to October 2025. Inefficient manual analysis potentially triggers overstocking due to a lack of understanding of consumer purchasing patterns. This research aims to analyze purchasing patterns using the FP-Growth algorithm to formulate operational recommendations. The analysis stages include data collection, preprocessing, data transformation, and the extraction of association rules. System evaluation was conducted by comparing manual calculations in Excel, Python output, and RapidMiner. This experiment utilized a minimum support parameter of 0.2% and a minimum confidence of 60%. The research results identified product association patterns, where one of the strongest rules indicates: if consumers buy Terigu Kompas 1Kg and Aqua 1500ml, they will also buy Terigu Kompas 500G (support 0.2%, confidence 60%, and lift ratio 21.95). Practically, this highly correlated figure provides a direct contribution to the store in the form of recommendations for placing these products adjacent to each other in the same aisle, as well as implementing bundling promotion strategies to minimize stock accumulation.

WEBSITE INFORMASI PARIWISATA DAERAH KABUPATEN MANGGARAI DENGAN SISTEM REKOMENDASI DESTINASI BERDASARKAN RATING

Vigo Angkur Vigo, Try Ana Setyarini Rini, Hasibun Asikin Hasibun — Thu, 10 Sep 2026 00:00:00 +0700

Manggarai Regency possesses significant tourism potential, yet remains constrained by the lack of integrated digital information, which hinders tourists from accessing accurate data and determining visit priorities. This research aims to design and develop a tourism information website integrated with an Item-Based Collaborative Filtering recommendation system to facilitate data-driven decision-making. The development employs the Research and Development (R&D) method using the Waterfall model, encompassing requirements analysis, system design, Implementation using PHP and MySQL, and rigorous functional testing. The final system was evaluated using Black Box testing, which confirms that all features, including the automated rating-based recommendation engine, function correctly according to specifications. The results demonstrate that the website effectively presents comprehensive information and generates accurate top-destination recommendations based on real-time visitor ratings. This platform provides a transparent and interactive tool that successfully assists tourists in selecting destinations while enhancing the digital promotion of Manggarai Regency’s tourism sector.

IMPLEMENTASI TRANSFER LEARNING DENGAN FINE-TUNING PADA DETEKSI OBJEK MULTI-KELAS MENGGUNAKAN YOLO (STUDI KASUS CAR FREE DAY JALAN EL TARI KOTA KUPANG)

Hendrikus Samuel Ola Sogen, Erna Rosani Nubatonis, Hasibun Asikin — Thu, 10 Sep 2026 00:00:00 +0700

Object detection is a computer vision technology used to recognize and determine the location of objects in images or videos. This study aims to implement the transfer learning method with fine-tuning on the YOLOv8m model to detect multi-class objects consisting of persons, vehicles, and umbrellas in the Car Free Day environment on El Tari Street, Kupang City, as well as to develop a web-based object detection system capable of automatically detecting objects in images and videos. The research dataset consisted of 315 images obtained through field documentation and annotated using Roboflow, which was then increased to 757 images through the augmentation process before being divided into training, validation, and testing datasets. The YOLOv8m pretrained model based on the COCO dataset was trained using Google Colab for 100 epochs with the transfer learning and fine-tuning approach. The results showed that the model achieved a precision of 0.903, a recall of 0.801, an mAP50 of 0.868, and an mAP50-95 of 0.625. In addition, the developed web-based system was able to automatically detect objects in images and videos and display bounding boxes, confidence scores, object counts, detection result graphs, and model evaluation metrics. The results of the study indicate that the application of fine-tuning to YOLOv8m is capable of improving the model's adaptability to the characteristics of the local Car Free Day environment, thereby potentially supporting more effective public activity monitoring.

ANALISIS PEMANTAUAN KUALITAS BBM ECERAN BERBASIS IOT DENGAN FUEL QUALITY SENSOR DAN SVM UNTUK MENENTUKAN KELAYAKAN BERDASARKAN SIFAT FISIK BBM

Gabrieno Bunyu, Menhya Snae, Heni — Thu, 10 Sep 2026 00:00:00 +0700

This research aims to design and develop an Internet of Things (IoT)-based monitoring system for the quality of retail petroleum products capable of operating in real time, in order to address the public’s limitations in independently verifying the suitability of retail petroleum products. The system was developed using a proximity sensor as an initial trigger to detect the presence of objects or liquids, and a TDS sensor as a fuel quality sensor to measure changes in conductivity values indicating the water content in fuel samples; each reading is locked for 5 seconds to ensure the stability of the proximity and TDS sensor values before the data is sent to a Flask-based server and analysed using a Support Vector Machine (SVM) algorithm to classify the fuel condition into the categories ‘Suitable’ and ‘Unsuitable’. Tests were carried out on 100 retail fuel samples, comprising 50 samples of pure fuel and 50 samples of fuel mixed with water, with classification rules based on TDS values: a value close to or equal to zero indicates that the fuel shows no signs of water contamination (Acceptable), whilst a value above zero which in the tests varied from 1 to over 700 depending on the level of contamination indicates the presence of water admixture (Unfit). The research results show that the system is capable of performing sensor readings, data transmission and the classification process effectively in real time; furthermore, based on an evaluation using a confusion matrix, the SVM model achieved an accuracy of 0.94, a precision of 0.946, a recall of 0.94 and an F1-score of 0.94. With this performance, the developed system has the potential to be utilised by the public and retail fuel businesses to verify fuel suitability quickly, automatically and objectively without the need for laboratory testing.

SISTEM IOT MONITORING SUHU MESIN DAN JARAK TEMPUH PADA SEPEDA MOTOR INJEKSI 4 - TAK DENGAN NOTIFIKASI PERAWATAN BERKALA

Esi Apriani Bunga, Menhya Snae, Heni — Thu, 10 Sep 2026 00:00:00 +0700

Motorcycles are the most widely used mode of transportation for people in their daily activities. Although modern 4-stroke injection motorcycles are equipped with temperature indicators and service reminders, these features are still very simple and cannot provide detailed and real-time vehicle condition information. Previous research generally only focuses on engine temperature monitoring or oil change reminders based on mileage, so they have not integrated both functions in a single Internet of Things (IoT)-based system. Based on these conditions, this study aims to design and implement an Internet of Things (IoT)-based system that integrates engine temperature and mileage monitoring and provides automatic periodic maintenance notifications. The system uses a DS18B20 sensor to monitor engine temperature, an A3144 Hall Effect sensor to calculate mileage based on wheel rotation, an ESP32 microcontroller as the main controller, and a Telegram Bot as a medium for sending notifications via a Wi-Fi network. Functional testing was carried out on engine temperature readings, mileage calculations, ESP32 connectivity to the Wi-Fi network, the LCD display, and sending Telegram notifications. The test results showed that all system components functioned according to the design. The DS18B20 sensor successfully monitors engine temperature and sends a Telegram notification when the temperature reaches ≥90°C, then sends a reminder every 2 minutes as long as the temperature remains at or above 90°C, and stops the notification when the temperature drops below 90°C. The A3144 Hall Effect sensor successfully calculates mileage based on wheel rotation, while the system automatically sends an oil change notification when the mileage reaches 2,000 km.

ANALISIS AKTOR PENENTU DAN PREDIKSI JENIS KONTRASEPSI PADA AKSEPTOR KB MENGGUNAKAN ALGORITMA RANDOM FOREST

Mario Edmon Gasa, Tri Ana Setyarini, Dewi Anggraini — Thu, 10 Sep 2026 00:00:00 +0700

The Family Planning (KB) program aims to control population growth, yet the high discontinuation rate due to mismatched contraceptive choices remains a major challenge in the field. Therefore, this study aims to develop an objective contraceptive prediction model using the Random Forest algorithm to minimize the risk of program failure. The methodology involved processing 4,500 acceptor records balanced into 9 contraceptive classes with 12 demographic variables, optimized via GridSearchCV, and evaluated using 5-Fold Cross Validation. The results indicate that the model operates stably with an average accuracy of 78.87%, achieving the best performance in the Fold-1 test at 81.67%. The model also demonstrated optimal recognition for the MOP and MAL classes (F1-Score 0.98), proving the algorithm's reliability in identifying classes with highly distinctive characteristics despite data overlap challenges within the Injectable and Pill classes. Feature Importance analysis reveals that Age (22.60%), Gender (14.39%), and Age at Marriage (12.44%) are the most dominant determining factors. This prediction model is implemented in a Flask application, serving as a practical decision-support tool for healthcare workers to provide instant, transparent, and targeted contraceptive recommendations.

ANALISIS SYSTEM TRACKING DAN KEAMANAN KENDARAAN BERMOTOR MENGGUNAKAN ARDUINO DAN ESP32

William Sanro Umbu Muda, Mehya Snae, Yohanis Malelak — Thu, 10 Sep 2026 00:00:00 +0700

Motor vehicle theft remains a leading criminal offense in Indonesia, highlighting the high vulnerability of factory-standard security systems. This situation underscores the urgent need for more adaptive technological innovations to enhance security and enable real-time vehicle tracking. This research focuses on developing an Internet of Things (IoT)-based vehicle security system integrated with location tracking capabilities. The study addresses the lack of an optimal comparative analysis regarding the performance of commonly used microcontrollers specifically the Arduino Uno and ESP32 within the context of tracking systems. System performance is evaluated based on three key parameters: GPS tracking accuracy (timestamp), data transmission latency (response time), and microcontroller power consumption efficiency. The findings aim to identify the superior and more efficient microcontroller platform, serving as a crucial reference for developers and researchers creating reliable, fast, and energy-efficient IoT-based vehicle security systems. The results indicate that the ESP32 outperforms the Arduino Uno R3; although the ESP32 exhibits higher power consumption and slightly lower location accuracy than the Arduino Uno R3, it offers a significant advantage: the ability to manage the SIM7600 module's functions and transmit data to the Blynk dashboard online capabilities that the Arduino Uno R3 lacks.

IMPLEMENTASI SISTEM PEMANTAUAN LEVEL AIR DAN ARUS LISTRIK POMPA SUMUR BOR MENGGUNAKAN HC-SR04 DAN ACS712

Adhy Syaputra Nge, Petrus Katemba, Heni — Thu, 10 Sep 2026 00:00:00 +0700

Access to clean water is a vital necessity for communities, particularly in areas that rely on borehole wells as their primary water source. However, manually operating these pumps without an adequate monitoring system can lead to issues such as dry-running, energy waste, and pump damage caused by electrical current irregularities. This study aims to implement an Internet of Things (IoT)-based monitoring system for borehole pump water levels and electrical current using an ESP32 microcontroller. Unlike systems that focus solely on water availability, this study integrates water level monitoring (via an HC-SR04 sensor), electrical current monitoring (via an ACS712 sensor), and a low-water protection mechanism (using a float switch) into a unified system featuring real-time Telegram notifications. The research methodology encompasses a literature review, field observations, system design, hardware and software implementation, and functional testing of each system component. Test results demonstrate that the HC-SR04 sensor successfully detects water level changes, the ACS712 sensor accurately reads the pump's electrical current, and the system effectively triggers automatic protection via a relay when low water levels are detected. Additionally, the system activates a buzzer alarm and sends warning notifications to the user via Telegram. Functional testing confirms that all system components operate according to the design, offering a practical solution to enhance the safety and reliability of borehole pump operations through real-time remote monitoring.

PREDIKSI PENJUALAN KEBAB BERDASARKAN POLA HARI DALAM SEMINGGU MENGGUNAKAN METODE SARIMA

Intan Maulida Hanum, Rudi Hariyanto, Muhammad Udin — Thu, 10 Sep 2026 00:00:00 +0700

Uncertainty regarding daily sales volume poses a challenge for inventory management in the culinary business, as it can lead to a mismatch between stock levels and demand. This study aims to apply the Seasonal Autoregressive Integrated Moving Average (SARIMA) method to forecast daily kebab sales using historical data from January 4, 2025, to January 3, 2026. The research process involved data preprocessing, splitting the data into training and testing sets, testing for stationarity using the Augmented Dickey-Fuller (ADF) test, identifying parameters via Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF) plots, and selecting the best model based on the Akaike Information Criterion (AIC) value. The ADF test results yielded a p-value of <0.05, indicating that the data was stationary. The optimal model identified was SARIMA (1,0,1)(1,0,1)₇, with an AIC value of 2509.27. Model evaluation resulted in an MAE of 15.09 portions, an RMSE of 17.09 portions, and a MAPE of 48.52%; these figures indicate that the average prediction error remains relatively high due to daily sales fluctuations. The model predicts sales of 26–28 portions per day, making it a useful reference for determining daily production volumes and raw material inventory requirements.

IMPLEMENTASI KEAMANAN RUMAH PINTAR BERBASIS IOT DENGAN ESP32, SENSOR GERAK, KAMERA DAN NOTIFIKASI REAL-TIME MELALUI BLYNK & TELEGRAM

Yosua Miha, Yohanes, Tri Ana Setyariny — Thu, 10 Sep 2026 00:00:00 +0700

This study addresses the slow response of conventional home security systems, which typically only record footage without providing immediate alerts during suspicious activity. The system utilizes ESP32 and ESP32-CAM microcontrollers as the core units, alongside a Passive Infrared (PIR) sensor, magnetic door sensor, MQ-2 smoke sensor, DHT22 temperature sensor, and vibration sensor for detection. The research methodology encompasses hardware design, software development using the Arduino IDE, integration with the Blynk and Telegram platforms, and functional testing of all system components. Implementation results confirm the successful construction of the system and the effective integration of all components. Testing demonstrated that the PIR sensor detects motion up to 5 meters away, triggering the ESP32-CAM to capture an image (with a 4 second delay) and send a notification containing the image to Telegram (with a delay of less than 1 second). The magnetic door sensor operates without delay when the door closes but exhibits a 1.5 second delay when the door opens, while notifications to Blynk are sent instantly. The MQ-2 sensor registers values below 200 in the absence of smoke and consistently exceeds 200 upon smoke detection, triggering a notification to Blynk. The DHT22 sensor does not send notifications at normal temperatures (28°C) but triggers an alert when the temperature exceeds 30°C. The vibration sensor sends notifications only when vibration is detected. All sensor data is processed by the ESP32 and transmitted to Blynk and Telegram via Wi-Fi. This research enhances security and environmental monitoring for weapon storage facilities, improving oversight through real-time integration with Blynk and Telegram

PENERAPAN METODE K-MEANS CLUSTERING DAN SUPPORT VECTOR MACHINE (SVM) BERBASIS MODEL RFM UNTUK KLASIFIKASI TIER PELANGGAN

Tariq, Nurmalitasari, Faulinda Ely Nastiti — Thu, 10 Sep 2026 00:00:00 +0700

Suboptimal management of large-scale transaction data can lead to marketing inefficiencies, particularly in determining promotional strategies that do not align with customer characteristics. This study aims to map the customer loyalty of CV Ekasa's client partners, by segmenting its customers using an integrated Recency, Frequency, Monetary (RFM) model, K-Means Clustering, and Support Vector Machine (SVM) classification. The dataset comprises 287,512 raw point-of-sale transaction records collected between October 2022 and September 2025, which after preprocessing yielded 341 valid customers for RFM modeling. Following the Cross-Industry Standard Process for Data Mining (CRISP-DM) framework, RFM features were log-transformed and standardized before clustering. Silhouette Score evaluation across k = 1–10 identified two customer segments (k = 2, Silhouette Score = 0.482) as optimal, labeled Passive Tier and Active Tier. These cluster labels were then used as classification targets for a linear-kernel SVM, evaluated under two data-splitting scenarios (80:20 and 70:30). The model achieved 97.10% accuracy with the 80:20 split and 98.06% with the 70:30 split, with precision, recall, and F1-scores above 0.97 for both tiers in both scenarios. These findings indicate that the integrated RFM–K-Means–SVM pipeline classifies customer loyalty tiers reliably and stably. The resulting model was deployed as an interactive Streamlit dashboard, giving CV Ekasa's client partner a practical, data-driven basis for designing more targeted and efficient marketing and retention strategies.

IMPLEMENTASI AUGMENTED REALITY SEBAGAI MEDIA PERAGA INTERAKTIF PADA PEMBELAJARAN HUKUM MEKANIKA FLUIDA

Matthew Chandra, Yohanis Malelak, Heni — Thu, 10 Sep 2026 00:00:00 +0700

Physics learning on Fluid Mechanics at State High School 1 Kupang faces obstacles due to students' difficulty in understanding abstract concepts using conventional methods. This study aims to design and implement an interactive Augmented Reality (AR) application on the Android platform to visualize 3D fluid mechanics objects. The development uses the Waterfall model, comprising needs analysis, UML design, implementation with Unity 3D and the Vuforia SDK (Marker-Based Tracking), and testing. System testing utilized Black Box testing for functionality and User Acceptance Test (UAT) questionnaires for user response, analyzed using descriptive quantitative methods on a sample of 30 students (n=30). The Black Box test yielded 100% validity without errors. Meanwhile, the UAT resulted in a feasibility score of 79.59%, categorized as "Good/Feasible". This research successfully produced a viable and innovative learning media application capable of increasing interactivity and students' conceptual understanding of Fluid Mechanics

KLASIFIKASI SURAT MASUK DI KANTOR PENGADILAN MILITER III-15 KUPANG MENGGUNAKAN (LSTM) LONG SHORT TERM-MEMORY

Asmawati Tuto, Sumarlin, Yohanis Malelak — Thu, 10 Sep 2026 00:00:00 +0700

The increasing volume of incoming correspondence at the Kupang Military Court III-15 Office has made the conventional letter classification process complex and time-consuming. Purpose: this study aims to develop an incoming letter classification system that categorizes letters into four classes (regular letters, circulars, decrees, and orders) using the Long Short-Term Memory (LSTM) method; the main contribution of this study is the application of LSTM to a local military-court correspondence dataset that has not been widely studied, together with a replicable preprocessing pipeline and K-Fold evaluation protocol, providing practical implications for accelerating correspondence administration in military judicial offices. Methods: the dataset consists of 500 incoming letter records in Excel format that underwent a preprocessing stage including cleaning, case folding, normalization, stopword removal, stemming, tokenizing, encoding, and padding, and was then divided into 80% training data and 20% testing data, evaluated using 5-Fold Cross Validation with accuracy, precision, recall, and F1-score as performance metrics. Results: the average model performance results were Accuracy 42.04%, Precision 42.39%, Recall 42.04%, and F1-Score 39.93%, with the highest accuracy obtained in Fold 2 (56.44%) and the lowest in Fold 5 (32.00%), while the model's main difficulty lay in distinguishing between the Circular and Decree categories, which share similar text patterns. Conclusion: the LSTM method is capable of recognizing textual patterns in letters and performing classification; however, its performance remains variable and relatively low on this dataset, so the developed system has the potential to improve the efficiency of incoming letter management at the Kupang Military Court III-15 Office, although further optimization is still needed before independent deployment.

SISTEM PELAPORAN PERBAIKAN DATA ALUMNI BERBASIS MOBILE WEB PADA UNIVERSITAS ARYASATYA DEO MURI

Lepri Veronika Nino, Erna Rosani Nubatonis, Heni — Thu, 10 Sep 2026 00:00:00 +0700

Accurate and up-to-date alumni data is an important aspect in supporting academic reporting and information management in higher education. However, the alumni data improvement process at Aryasatya Deo Muri University is still done manually so it is less efficient, takes longer time, and makes it difficult for alumni to submit and monitor the status of data improvement. This study aims to design and build a mobile web-based Alumni Data Improvement Reporting System to facilitate alumni, academic departments, and PDDIKTI operators in the process of submitting, verifying, and monitoring data improvement. The system was developed using the Waterfall method which includes needs analysis, design, implementation, testing, and maintenance, by utilizing PHP, MySQL, and Bootstrap technology. System testing was carried out using the Black Box method and user evaluation through questionnaires to 57 alumni respondents selected using the Slovin formula. The results showed that all system functions ran according to needs, while the results of the user evaluation obtained a satisfaction level of 80.79% with the Strongly Agree category, which indicates that the system is easy to use and able to support the alumni data improvement process effectively. The main contribution of this research is the availability of a system that integrates supporting document upload features, verification and validation processes, and real-time monitoring of application status, thereby increasing the efficiency of alumni data management and supporting the provision of more accurate, up-to-date, and well-documented data.

PENERAPAN REGRESI LINEAR BERGANDA DALAM MEMPREDIKSI POPULASI SAPI DI NUSA TENGGARA TIMUR

Rithwan Bernadus Rissi, Sumarlin, Yohanis Malelak — Thu, 10 Sep 2026 00:00:00 +0700

Nusa Tenggara is one of the largest beef cattle-producing provinces in Indonesia; however, the dynamics of the cattle population in this region show unstable fluctuations, including a sharp decline of 53.2% from 1,243,884 head (2022) to 581,918 head (2023). This study aims to analyze the influence of rainfall, feed availability, and beef production on the beef cattle population in East Nusa Tenggara, while also developing a prediction model based on multiple linear regression. Secondary data for the 2014–2024 period were obtained from the East Nusa Tenggara Central Statistics Agency. Missing values were imputed using quadratic interpolation and linear extrapolation. The regression model was estimated using Cramer’s rule and implemented in a Python-based prediction system with a Tkinter graphical interface. The research results yielded the regression equation with a coefficient of determination R² = 0.570473, MAE = 98,348.3527 head, MSE = 13,841,152,103.80, and RMSE = 117,648.43 head. The predicted cattle population for 2025 is 628,276.38 head, indicating a recovery trend. This model serves as an initial tool for local governments in planning strategies for livestock population recovery in East Nusa Tenggara.

PENERAPAN WEBGIS UNTUK VISUALISASI DAN ANALISIS LOKASI SEKOLAH DI KECAMATAN NUBATUKAN KABUPATEN LEMBATA

Djunus Bruno Djunior, Remerta N. Na’atonis, Skolastika Siba Igon — Thu, 10 Sep 2026 00:00:00 +0700

Nubatukan District, Lembata Regency, has 42 schools ranging from elementary to senior high school level spread across its villages, yet information on their locations and distribution has not been presented spatially in an accessible way. This study aims to design and build a WebGIS application as a medium for visualizing and managing school location data in the district. The system was developed using PHP, HTML, CSS, JavaScript, and a MySQL database, supported by ArcGIS and Google Earth Pro for spatial data processing, and was tested through blackbox testing and a Likert-scale User Acceptance Test (UAT) involving 44 respondents. Blackbox testing results show that all application functions operate according to the designed requirements, while the UAT obtained a user acceptance rate of 89.13%, categorized as very good. The developed WebGIS application therefore effectively displays the distribution of schools interactively and benefits the community and local government of Nubatukan District in accessing educational information.

ANALISIS SISTEM MONITORING TEKANAN BAN REAL TIME BERBASIS ESP32 DENGAN SENSOR TPMS PADA KENDARAAN BERMOTOR RODA DUA

Albert Valentino Mata Rohi, Petrus Katemba, Semlinda Juszandri Bulan Bulan — Thu, 10 Sep 2026 00:00:00 +0700

Tire pressure is a critical factor affecting vehicle safety, riding comfort, fuel efficiency, and tire lifespan. Conventional tire pressure inspection is generally performed manually, making continuous monitoring difficult. This study aims to develop a real-time Tire Pressure Monitoring System (TPMS) based on the ESP32 microcontroller integrated with the Blynk Internet of Things (IoT) platform for monitoring tire pressure and temperature on two-wheeled motor vehicles. The system was developed using the prototype method, including requirement analysis, hardware and software design, implementation, and performance evaluation. Functional testing was conducted by comparing TPMS pressure readings with a standard tire pressure gauge and measuring the transmission latency from the ESP32 to the Blynk application under various operating conditions. The experimental results show that the proposed system achieved an average front tire pressure measurement error of 1.08% while rear wheel error averaged at 0.44% and an average data transmission delay of 0.431 millisecond, indicating that the system provides accurate measurements and fast real-time communication. Furthermore, the developed system successfully monitored tire pressure and temperature continuously, enabling early detection of abnormal tire conditions and supporting preventive maintenance. These findings demonstrate that the proposed IoT-based TPMS offers a practical, low-cost, and reliable solution for improving motorcycle safety through continuous tire condition monitoring.

IMPLEMENTASI AUGMENTED REALITY SEBAGAI MEDIA PEMBELAJARAN TEKNIK DASAR PADA CABANG OLAHRAGA SHORINJI KEMPO

Putri Sezilia Kos, Remerta Noni Na’atonis, Skolastika Siba Igon — Thu, 10 Sep 2026 00:00:00 +0700

This study aimed to develop an Augmented Reality (AR) application as a learning tool for basic Shorinji Kempo techniques, helping kenshi (practitioners) understand and practice these techniques in a more interactive manner. The study employed the Research and Development (R&D) method using the ADDIE model, which comprises the stages of Analysis, Design, Development, Implementation, and Evaluation. The application was developed using Unity 3D and the Vuforia SDK, while testing involved 17 respondents selected from a population of 18 kenshi using Slovin's formula. Evaluation was conducted through Black Box testing to verify application functionality and a Likert-scale questionnaire to assess the application's suitability. The results demonstrated that the application performed well on Android devices, all functions operated as intended, and it achieved a suitability rating of 83.34%, placing it in the "Highly Suitable" category. The findings indicate that the Augmented Reality application is suitable for use as a learning tool for basic Shorinji Kempo techniques and can serve as a supporting resource for both dojo training and independent study

IMPLEMENTASI AUGMENTED REALITY SEBAGAI MEDIA VISUALISASI STRATEGI LATIHAN SEPAK BOLA PADA KLUB LOKAL AS RAEMANA

Nadio Rodriques Gili Raga, Dewi, Yohanis — Thu, 10 Sep 2026 00:00:00 +0700

The development of Augmented Reality (AR) technology provides significant opportunities to improve the quality of sports training, particularly in soccer, through more interactive and realistic strategy visualization. This study aims to design and develop an AR application as a visualization medium for soccer training strategies at the local AS Raemana soccer club, which faces limitations in strategy delivery through conventional methods such as whiteboards and verbal instructions, making it difficult for players to understand formations and game patterns in attacking and defensive situations. This study employed the Research and Development (R&D) method consisting of assessment, needs assessment, front-end analysis, design, development, implementation, and evaluation stages. The application was developed using Unity Engine and Vuforia SDK for Android, with key features including formation selection, strategy selection, marker scanning, and real-time 3D object visualization. Functional testing results showed that all application features operated according to the design and were declared valid. Based on a questionnaire administered to 21 respondents consisting of coaches and players of AS Raemana, all aspects received positive responses, with the percentage of respondents selecting “Strongly Agree” ranging from 61.9% to 100%, while the remaining respondents selected “Agree.” These results indicate that the application can help coaches deliver strategies more clearly and assist players in understanding formations, positions, movements, and game strategies in both attacking and defensive situations. This study contributes an interactive and innovative AR-based training medium that has the potential to support the improvement of tactical understanding and training quality in local soccer clubs.

IMPLEMENTASI SISTEM SKIRINING AUTISM SPECTRUM DISORDER BERBASIS WEB MENGGUNAKAN SUPPORT VECTOR MACHINE YANG DI OPTIMALKAN DENGAN HYPERPARAMETER TUNING

Adam Galuh Bhakti, Brestina Gultom — Thu, 10 Sep 2026 00:00:00 +0700

Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder that affects communication, social interaction, and behavior in children. Early detection of ASD is crucial because early intervention has been proven to improve children's adaptive abilities. However, the screening process in Indonesia is still conducted conventionally through observation and interviews, which require considerable time and are highly subjective. This study aims to implement a prototype web-based ASD screening system using a Support Vector Machine (SVM) optimized through feature selection and hyperparameter tuning. The dataset used is the Autism Child Screening Data (n=292) with 22 initial variables covering demographic data and Q-CHAT-10 screening results. The class distribution is relatively balanced, with 151 Normal (51.7%) and 141 Autism (48.3%) samples. The research stages include data preprocessing, feature selection using four methods (correlation, Mutual Information, ANOVA F-score, and Permutation Importance), hyperparameter tuning with GridSearchCV 10-fold cross-validation, model training, evaluation, and web-based system implementation using Streamlit. The results show that the SVM model with the RBF kernel, parameters C=10.0, gamma=0.01, and class weight {0:1.0, 1:3.0}, achieved an accuracy of 94.83%, a precision of 100% for the Normal class and 90.32% for the Autism class, a recall of 90% for Normal and 100% for Autism, an F1-score of 0.95 for both classes, and an ROC-AUC of 0.9940. The model is implemented in a web prototype that can be used as an early screening aid for healthcare professionals. This research contributes to the development of a machine learning-based decision support system integrated with clinical knowledge through the Q-CHAT-10 cut-off post-processing rule.

KLASIFIKASI SENTIMEN MASYARAKAT TERHADAP AKSI 17+8 DI MEDIA SOSIAL MENGGUNAKAN LSTM DAN BERT

Sonia Roselina Correia, Sumarlin, Dewi Anggraini — Thu, 10 Sep 2026 00:00:00 +0700

This study aims to compare the performance of Long Short-Term Memory (LSTM) and Bidirectional Encoder Representations from Transformers (BERT) models in classifying public sentiment on platform X (Twitter) regarding the "17+8 Tuntutan Rakyat" movement. The dataset consists of 1,000 Indonesian tweets collected between August and September 2025, with a subset of 200 data evaluated using a 5-Fold Cross Validation scheme. The average evaluation results show that the LSTM model achieved an accuracy of 0.6900, whereas BERT achieved 0.5400. However, per-class metric analysis reveals that LSTM suffered from severe majority-class bias by predicting all instances as neutral (F1-score of 0.0000 for both positive and negative classes), whereas BERT demonstrated discrimination capability on minority classes (negative recall of 26.83% and positive recall of 14.29%). This research is limited by a small evaluation subset size and the absence of class imbalance handling techniques, which implies the crucial need for GPU acceleration and resampling methods in future social media text sentiment analysis studies.

PENGEMBANGAN SISTEM INVOICE DIGITAL DENGAN FITUR AUDIT TRAIL UNTUK PELACAKAN TRANSAKSI MENGGUNAKAN PENGUJIAN USABILITAS PENGGUNA

Aang Samsudin Samsudin, T Tara Wanesha — Thu, 10 Sep 2026 00:00:00 +0700

PT. XYZ is a manufacturing industry supplier that uses invoices as official billing documents. However, invoice management is currently handled via Microsoft Excel, leading to potential recording errors, difficulties in data retrieval, and suboptimal monitoring of installment-based payments. Therefore, this study aims to design a website-based digital invoicing system to automate the recording process and provide payment status tracking features. The study employs a prototyping development method, enabling direct user involvement throughout the design process encompassing communication, quick planning and design, prototype construction, and delivery and feedback stages. The system is built using the Laravel framework to ensure a structured, secure, and scalable system tailored to the company's needs. Testing was conducted using the black-box method across 15 functional and interface scenarios covering input, processing, and invoice reporting. Results indicate an 88.67% success rate, with an average response time of under two seconds and a 5% error rate for key features, classifying the application as highly viable. Implementing this system offers practical benefits by improving operational efficiency in financial data reconciliation compared to the previous system.

ANALISIS DAN REDESAIN PROSES BISNIS PEMBELIAN BERBASIS BPM PADA TOKO RITEL EMAS

Riyan Ainur Rahman, Nur Aeni Hidayah, Maulana Asykari Muhammad — Thu, 10 Sep 2026 00:00:00 +0700

The purchasing business process at Toko Emas Asia Ceger still experiences operational inefficiencies caused by sequential activities after invoice generation, resulting in waiting time during inventory updates and delayed inventory recording. This study aims to analyze the existing purchasing business process (AS-IS) and develop a redesigned business process model (TO-BE). A qualitative approach was employed through three days of non-participant observation and semi-structured interviews involving three operational staff representing the Sales, Finance, and Inventory divisions. The analysis integrated Business Process Model and Notation (BPMN) 2.0, Value Added Analysis (VAA), Root Cause Analysis (RCA) using a Fishbone Diagram, and Business Process Redesign (BPR). The findings indicate that the main source of inefficiency is the sequential dependency between payment, product delivery, and inventory updates. Based on process-based estimation, the proposed TO-BE model introduces a parallel gateway after invoice generation, reducing estimated waiting time in the Inventory Division from approximately 8 minutes to 2 minutes (75%) and overall process lead time from approximately 28 minutes to 22 minutes (21%). This study contributes an integrated BPMN–VAA–RCA–BPR evaluation framework for analyzing purchasing business processes in local gold retail and provides a TO-BE business process model as a basis for process improvement.

AUGMENTED REALITY SEBAGAI MEDIA EDUKATIF INTERAKTI DALAM PENGENALAN PAKAIAN ADAT LEMBATA PADA SISWA SEKOLAH DASAR INPRES DULITUKAN

Marcello Bernado S. Balawala, Remerta Noni Na’aTonis, Petrus Katemba — Thu, 10 Sep 2026 00:00:00 +0700

This study aims to design and develop an Augmented Reality (AR) application as an interactive learning medium to introduce Lembata traditional clothing to students of SD Inpres Dulitukan. Through AR technology, traditional clothing objects are displayed in three-dimensional form that can be observed directly through digital devices, making the learning process more interesting, easier to understand, and not limited to conventional visual materials. In addition, the application is equipped with audio explanations regarding the meaning and characteristics of each traditional clothing, providing a more comprehensive and contextual learning experience. The research method used is the Multimedia Development Life Cycle (MDLC), which consists of data collection, analysis, design, media development, testing, and evaluation stages. This method was chosen because it is suitable for multimedia-based application development and allows the development process to be carried out systematically. The results of the study indicate that the developed Augmented Reality application runs well on Android devices and is capable of displaying Lembata traditional clothing objects in three-dimensional form along with automatic audio explanations. Based on the Black Box testing results, all application features functioned according to the design. In addition, the questionnaire results showed that students gave positive responses to the application because it was considered interesting, easy to use, and helpful in understanding the introduction to Lembata traditional clothing. Therefore, this application can be used as an interactive educational medium to support local cultural learning while also helping preserve regional cultural heritage through the application of modern technology.

KLASIFIKASI KELAYAKAN PRODUK PANGAN UMKM MENGGUNAKAN MACHINE LEARNING UNTUK MENDUKUNG PROGRAM MAKAN BERGIZI GRATIS

Ainun Hidayah, Ismail Ismail, Ghina Raudhatul Janna, Ikra Juwita, Nur Khalizah — Thu, 10 Sep 2026 00:00:00 +0700

The Free Nutritious Meal Program requires an objective and standardized approach to evaluate the eligibility of food products supplied by Micro, Small, and Medium Enterprises (MSMEs). This study aims to develop a machine learning-based classification model using Random Forest to support the initial screening of MSME food product eligibility. A real-world dataset containing 120 MSME food product records was utilized, consisting of nutritional, economic, legality, certification, and packaging quality attributes. The data were preprocessed and divided into training and testing sets using an 80:20 ratio. The Random Forest model achieved the best classification performance, obtaining an accuracy of 96.00%, precision of 93.75%, recall of 100.00%, and F1-score of 96.77%. Feature analysis showed that PIRT license, packaging hygiene, and halal certification were the most influential factors in determining product eligibility. The proposed model provides practical support for improving the objectivity, efficiency, and documentation of MSME food product screening in the implementation of the Free Nutritious Meal Program. This study is limited by the use of a small-scale prototype dataset; therefore, future research should involve larger real-world datasets and further validation in operational environments.

KLASIFIKASI SPASIAL TINGKAT KEMISKINAN RUMAH TANGGA MENGGUNAKAN NAIVE BAYES DI KECAMATAN WOTAN ULUMADO

Cyrilius Budi De Fe Rento Lewo Manuk, Meliana O.Meo, Tri Ana Setyarini — Thu, 10 Sep 2026 00:00:00 +0700

Household poverty is a social issue that requires accurate identification to ensure that poverty alleviation programs are implemented effectively and targeted appropriately. Wotan Ulumado Subdistrict, East Flores Regency, has diverse socioeconomic characteristics; therefore, a data-driven method is needed to classify household poverty levels. This study aims to analyze the socioeconomic factors associated with household poverty and apply the Naive Bayes method to classify households according to their poverty status. The data were collected through field observations, interviews, and a literature review. The variables examined included household income, number of dependents, the educational level of the household head, housing conditions, and asset ownership. The dataset consisted of 360 training records and 28 testing records. The classification process was conducted by calculating the prior, likelihood, and posterior probabilities for each poverty category. The classification categories comprised poor, moderately poor, and non-poor households. Evaluation using a confusion matrix showed that 26 out of 28 testing records were correctly classified, resulting in an accuracy rate of 92.86%. These findings indicate that the Naive Bayes method performs well in identifying household poverty levels based on socioeconomic indicators. The classification results can further be presented in the form of tables, graphs, and maps to illustrate the distribution of poverty levels and support local government decision-making in determining priority households and areas for poverty alleviation programs in Wotan Ulumado Subdistrict.

EVALUASI SISTEM PEMERINTAHAN BERBASIS ELEKTRONIK (SPBE) DOMAIN LAYANAN PUBLIK PADA DINAS PARIWISATA KABUPATEN KUPANG

Kresensia B. Hewen, Tri Ana Setyarini, Meliana O. Meo — Thu, 10 Sep 2026 00:00:00 +0700

The implementation of the Electronic-Based Government System (SPBE) has become an important strategy for improving the quality, efficiency, transparency, and accountability of public services in Indonesia. However, the effectiveness of SPBE implementation depends on the maturity of its governance and the integration of digital services across government institutions. This study aims to evaluate the implementation of the SPBE public service domain at the Kupang Regency Tourism Office based on the SPBE maturity assessment framework. A mixed-methods approach was employed by integrating quantitative and qualitative data. Quantitative data were collected through a 24-item questionnaire distributed to 66 respondents consisting of government employees, tourists, and tourism business actors, while qualitative data were obtained through interviews, observations, and documentation to support and explain the quantitative findings. The quantitative data were analyzed using a five-point Likert scale, whereas qualitative data were analyzed descriptively through data reduction, data presentation, and conclusion drawing. The results indicate that the SPBE public service domain achieved an index score of 3.17 (63.40%), categorized as Fair and corresponding to Maturity Level 3 (Defined). The highest score was achieved in electronic government administration services, while public complaint services, system interoperability, and service integration obtained the lowest scores. These findings suggest that SPBE implementation has been established but has not yet been fully integrated across the public service process. Therefore, strengthening system interoperability, improving electronic complaint services, and enhancing service integration are recommended as priority strategies to improve the quality, effectiveness, and sustainability of digital public services at the Kupang Regency Tourism Office.

PREDIKSI PENJUALAN OBAT PADA PEDAGANG BESAR FARMASI MENGGUNAKAN METODE PERBANDINGAN EKSPONENSIAL (MPE)

Fransiskus Masan, Sumarlin, Heni — Thu, 10 Sep 2026 00:00:00 +0700

Mathio Jaya Pharma is a pharmaceutical distribution company that plays an important role in maintaining drug inventory to meet customer demand. Inappropriate inventory management can lead to overstocking, which increases the risk of product expiration, or stock shortages that may disrupt customer service. This study aims to predict drug sales as a basis for inventory procurement planning for the following period. The methods used in this study are the Entropy Method to determine objective criterion weights based on data variation and the Exponential Comparison Method to generate drug sales predictions. The data used consist of drug sales records from January 1 to December 31, 2025, with inventory, price, and total sales as the evaluation criteria. The weights obtained from the Entropy Method are used as input for the MPE calculation to produce sales predictions for the next period. The results indicate that the system is capable of generating drug sales predictions and providing inventory status information categorized as safe or critical. Based on the prediction results, Paracetamol, Dexa, and Caviplex are classified as safe because the available stock is sufficient to meet future demand, while Amoxicillin is categorized as critical and requires additional procurement. Accuracy testing using the Mean Absolute Percentage Error (MAPE) produced values of 74.51% for Paracetamol, 58.33% for Caviplex, 47.51% for Dexa, and 33.33% for Amoxicillin. The findings indicate that the combination of the Entropy Method and the Exponential Comparison Method can assist the company in predicting future drug sales and support more effective and efficient decision-making in inventory management.

IMPLEMENTASI AUGMENTED REALITY (AR) SEBAGAI MEDIA PEMBELAJARAN INTERAKTIF DALAM VISUALISASI PROYEK INFRASTRUKTUR JEMBATAN

Kanisius Dominggus Mua, Yohanis, Hasibun — Thu, 10 Sep 2026 00:00:00 +0700

This study aims to develop an interactive learning medium based on Augmented Reality (AR) that is capable of visualizing bridge structures clearly, realistically, and comprehensively for students of SMKS Sanjaya Bajawa, as well as evaluating its effectiveness as a learning medium. The research employed the Research and Development (R&D) method using the Lee & Owens model, which consists of the stages of Assessment/Analysis, Needs Assessment, Front-End Analysis, Design, Development, Implementation, and Evaluation. The application was developed using Unity 3D, Vuforia SDK, and Blender to create a three-dimensional bridge model that can be displayed through Augmented Reality technology on Android devices. The results of the study indicate that the ARJembatan application was successfully developed and that all major features functioned properly based on Black Box Testing. The results of the User Acceptance Testing (UAT), involving 10 eleventh-grade students of SMKS Sanjaya Bajawa, showed that the majority of respondents provided positive evaluations of the application. As many as 90% of respondents stated that the application was beneficial as a learning medium, 90% considered the application menu easy to use, and 90% found the implementation of Augmented Reality technology in the application engaging and attractive. Furthermore, the interactive 3D visualization helped students understand bridge structures and components more clearly compared to conventional learning media.

ANALISIS PHISHING PADA PESAN WHATSAPP MENGGUNAKAN LSTM

Alison Dejesus, Petrus Katemba, Ernarosani Nubatonis — Thu, 10 Sep 2026 00:00:00 +0700

Phishing messages on instant messaging applications such as WhatsApp pose a cybersecurity threat, particularly through social engineering techniques, making automated detection of Indonesian-language phishing messages important. This study aims to develop a phishing detection system for Indonesian-language WhatsApp messages using the Long Short-Term Memory (LSTM) model. The dataset consisted of 1,162 messages, comprising 817 non-phishing and 345 phishing messages, which were processed through cleaning, case folding, normalization, tokenization, stopword removal, stemming using Sastrawi, sequencing, and padding before model training. The LSTM model was trained using 5 epochs, a batch size of 32, a learning rate of 0.001, and the Adam optimizer, and evaluation using a confusion matrix achieved an accuracy of 88.0%, precision of 72.50%, recall of 96.67%, and F1-score of 82.86%. These results indicate that the model can support the initial identification of potentially phishing messages, particularly due to its high recall in detecting phishing messages, although the imbalanced class distribution remains a limitation that may affect classification performance.