Bulletin of Information Technology (BIT)

Prototype Sistem Deteksi Penyakit Mulut dan Kuku Menggunakan Gambar Citra Digital Sebagai Upaya Menjaga Kesehatan Ternak di Kabupaten Sumbawa

Siska Atmawan Oktavia — 2025-12-14

The cattle industry worldwide faces a major threat from foot-and-mouth disease (FMD) due to the contagious nature of the virus. In Sumbawa Regency, FMD peaked in 2022 with 12,814 cases. Screening for FMD is very important for early detection and treatment. Currently, detection conducted by animal health officers is still manual, requiring 48 hours per animal to obtain diagnostic results and is prone to errors. The researchers aim to create a Prototype Automatic Detection System that includes an automatic foot-and-mouth disease (FMD) detection system application using App Designer (GUI) system-MATLAB to assist animal health officers in diagnosing animals infected with FMD, saving time and costs and saving animals. The proposed method for automatically extracting distinguishing features of cattle and classifying whether the cattle are sick or healthy utilizes the advantages of the Convolutional Neural Network (CNN) model. Based on the evaluation results of the developed system, the proposed system using the Convolutional Neural Network algorithm has better performance with an accuracy of 100% compared to the WEKA application, namely SMO with an accuracy of 90%, IBk (87%), Trees.J48 (86%), and Naive Bayes (79%). Therefore, highly efficient and accurate digital image processing techniques must be used to produce effective FMD disease screening. The proposed decision support system for clinical screening is expected to make a significant contribution and help reduce the workload of Animal Health Officers in detecting foot-and-mouth disease (FMD).

Rancang Bangun Media Edukasi 3D Interaktif Pembelajaran IPA Berbasis Mobile

Suharjanto Utomo — 2025-12-14

The rapid development of digital technology provides new opportunities to improve the quality of learning, particularly in science education. Conventional learning media such as textbooks and static images are considered less effective in conveying concepts that are highly visual and abstract. This study aims to design and develop an interactive 3D mobile-based educational media as a learning tool for science subjects, with a case study on the human digestive system at SMP Muhammadiyah 5 Cimahi. The application enables students to explore the digestive system organs through interactive 3D models supported by audio narration and explanatory texts. The development method employed is the Multimedia Development Life Cycle (MDLC), which consists of six stages: concept, design, material collecting, assembly, testing, and distribution. The application was evaluated through alpha testing using the blackbox method and beta testing by distributing questionnaires to teachers and 14 eighth-grade students. The evaluation results indicate that the application is well-accepted and contributes positively to students’ understanding of the human digestive system. The questionnaire results showed a score of 4.37, indicating that the application is engaging, easy to use, and interactive. These findings highlight the potential of interactive 3D mobile-based educational media as an innovative alternative for more engaging and effective science learning.

Comparative Analysis of ARIMA and LSTM Methods for Forecasting Healthcare Service Costs in Advanced Referral Healthcare Facilities in Bogor City

Rizal Rizal Ashari Nampira — 2025-12-14

The National Health Insurance (JKN) program, managed by BPJS Kesehatan, has experienced a significant increase in healthcare service costs, particularly at Advanced Referral Healthcare Facilities (FKRTL). This study aims to compare the forecasting accuracy of ARIMA and Long Short-Term Memory (LSTM) methods in predicting healthcare service costs in FKRTL Bogor from January 2014 to October 2024. The data, sourced from BPJS Kesehatan Branch Bogor, were analyzed using time series approaches. Model evaluation was conducted using Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). Results show that for 80% of training data, LSTM produced a MAPE of 8.85% and RMSE of IDR 6.98 billion, slightly outperforming ARIMA (0,1,1) with MAPE of 10.28% and RMSE of IDR 6.67 billion. For the 20% testing data, LSTM demonstrated significantly better accuracy, with an MAPE of 12.97% and RMSE of IDR 15.52 billion, compared to ARIMA’s MAPE of 24.22% and RMSE of IDR 30.76 billion. Therefore, LSTM is considered more effective for short- to medium-term forecasting of JKN healthcare costs, particularly when dealing with complex and non-linear patterns.

Clustering Academic Data of Junior High School Students to Identify Learning Groups Using The DBSCAN Algorithm at SMP Muhammadiyah 5 Samarinda

Mini H — 2025-12-14

The formation of study groups at the junior high school level plays an important role in improving the quality of learning and promoting equality in student learning outcomes. However, the process of grouping students is still largely carried out manually based on teachers’ intuition, subjective observations, or attendance data, which may lead to mismatches in students’ abilities and hinder the optimal achievement of learning objectives within the school environment. This study aims to identify study groups based on students’ academic data at SMP Muhammadiyah 5 Samarinda. The data used include scores in science (exact) and non-science (non-exact) subjects, exam results, assignment scores, attendance records, and parents’ educational backgrounds. The research stages consist of data cleaning, feature engineering, standardization, the application of the DBSCAN algorithm, and evaluation using the Silhouette Score. The analysis results reveal three main clusters: cluster 0 with 89 students (medium achievement), cluster 1 with 50 students (high achievement), and cluster 2 with 5 students (low achievement). In addition, 14 students (8.9%) were identified as noise. The Silhouette Score value of 0.217 indicates that the cluster separation quality is relatively weak; however, DBSCAN successfully detected outliers that may not be identified by other algorithms. These findings suggest that, although the cluster quality is not yet optimal, the applied algorithm remains useful for exploring students’ learning patterns and can serve as a basis for more targeted learning interventions.

Application of The Naïve Bayes Algorithm for Employee Performance Prediction Based on SIMPEG at TVRI East Kalimantan Station

Ishmah Hanani — 2025-12-14

Employee performance evaluation is a crucial aspect of public organizational management, including at the public broadcasting institution TVRI East Kalimantan Station. To date, attendance indicators obtained from the Employee Management Information System (SIMPEG) have often been used as the primary benchmark, as the data are objectively and structurally available. However, a single attendance-based approach risks overlooking more substantive aspects of work achievement. Therefore, this study integrates attendance data with the Employee Performance Targets (SKP) to construct a more representative performance label. The method employed is a classification approach using the Naïve Bayes (GaussianNB) algorithm. The research dataset consists of attendance records (normal attendance, leave, official duty, study assignment, early departure, absence, and total working days) and quantized SKP scores. Performance labels were generated using a composite score (0.30 × attendance percentage + 0.70 × normalized SKP), which was then categorized into three classes: Excellent, Good, and Needs Improvement. The model was trained using SIMPEG and SKP data that had undergone preprocessing, data partitioning, and class balancing. Experimental results show that the model achieved an accuracy of 0.83, with a precision of 0.86, recall of 0.84, and F1-score of 0.83 on the test data. These results indicate that the model can consistently recognize employee performance patterns across all categories. Practically, this study offers a simple, efficient, and easily implementable predictive framework to support more objective processes of coaching, monitoring, and reward allocation within TVRI East Kalimantan Station.

Penerapan Algoritma XGBoost Dalam Prediksi Harga Sewa Kos Di Kota Samarinda

Amalia Rahman — 2025-12-14

The population growth and increasing economic activity in Samarinda City have led to a rising demand for temporary housing such as boarding houses. However, rental price determination is still largely based on the owner’s intuition rather than objective factors such as available facilities, room specifications, transportation accessibility, and proximity to public amenities. This study aims to develop a rental price prediction model for boarding houses using the Extreme Gradient Boosting (XGBoost) algorithm with a Knowledge Discovery in Database (KDD) approach. The research data were collected through a web scraping process from the Mamikos platform, yielding 231 initial records, which were then cleaned and filtered for outliers, resulting in 225 valid data points. Five main features derived from feature engineering were utilized in the model, namely Facility Score, Combined Specification Score, Nearest Place Score, Transportation Score, and Rental System Score. The evaluation results show that the XGBoost model achieved a Mean Absolute Error (MAE) of Rp348,822, a Root Mean Squared Error (RMSE) of Rp416,139, and a coefficient of determination (R²) of 0.612. These values indicate that the model can explain 61.2% of the variation in rental prices with reasonably good predictive performance. The feature importance analysis reveals that Facility Score and Combined Specification Score are the most influential factors affecting rental prices, while transportation and rental system factors contribute less significantly. This study is expected to serve as a reference for boarding house owners, tenants, and policymakers in determining more objective and competitive rental prices based on a data mining approach.

Pengenalan Tarian Adat Dayak Hudoq Melalui Media Virtual Reality Untuk Pelestarian Budaya

Otovianus Oscar — 2025-12-14

Abstract- The preservation of the traditional Dayak Hudoq dance faces serious challenges due to globalization and the use of conventional learning media that are less appealing to the digital native generation. These challenges have resulted in a lack of understanding among the younger generation of the philosophical values of dance. There is a research gap in the use of Virtual Reality (VR) technology that focuses on the philosophical values of the dance and can be accessed independently (offline) using low-cost devices. This study aims to (1) develop an offline-based Virtual Reality educational application (VR-Box) using the Multimedia Development Life Cycle (MDLC) method, and (2) test the feasibility of this media as a means of cultural preservation. The MDLC method is applied through six systematic stages: Concept, Design, Material Collecting, Assembly, Testing, and Distribution. The testing process involved Alpha testing for functionality and Beta testing using a Likert scale questionnaire with 10 vocational high school students in Samarinda to measure ease of use and appeal. The results of the study show that the VR-Box application was successfully developed and functions well. The beta test results show a “Very Good” level of user acceptance with an overall average score of 88%. This application is considered very practical to use (96%) and capable of increasing interest in learning about culture (88%). It is concluded that the VR-Box application is feasible and effective to be implemented as a portable and low-cost medium for cultural preservation. However, user evaluation shows that visual quality (74%) is still an aspect that needs to be improved in further research.

Keywords: MDLC, Visualization, Dayak, Cultural Preservation, VR

Perbandingan Metode ARAS dan TOPSIS untuk Menentukan Siswa Penerima Beasiswa Berprestasi

Jeperson Hutahaean — 2025-12-14

Proses penentuan siswa yang layak menerima beasiswa berprestasi seringkali hanya berdasarkan seleksi berkas, yang memakan waktu lama dan dapat menghasilkan keputusan yang kurang akurat. Oleh karena itu, penelitian ini bertujuan untuk mengembangkan sistem pendukung keputusan yang mempercepat proses seleksi siswa penerima beasiswa berprestasi. Metode yang digunakan dalam penelitian ini adalah ARAS (Additive Ratio Assessment) dan TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution). Kedua metode ini dipilih karena mampu menyeleksi alternatif terbaik berdasarkan pembobotan dan memilih tujuan berdasarkan kriteria yang berbeda, yaitu benefit dan cost. Metode ARAS menghasilkan ranking berdasarkan perbandingan fungsi utilitas alternatif dengan nilai fungsi utilitas optimal, sementara metode TOPSIS menghasilkan ranking berdasarkan jarak terpendek dengan solusi ideal positif dan jarak terpanjang dengan solusi ideal negatif. Dari nilai-nilai tersebut, alternatif yang memenuhi kriteria ditetapkan sebagai siswa penerima beasiswa berprestasi melalui perhitungan menggunakan metode ARAS dan TOPSIS. Hasil uji korelasi Rank M Arsya Almusa menunjukkan nilai 1,0546 untuk ARAS dan 0,8739 untuk TOPSIS. Hasil perhitungan secara manual dan melalui sistem memberikan hasil yang sama, sehingga sistem dapat digunakan untuk menentukan siswa penerima beasiswa berprestasi.

Penerapan Data Mining Menggunakan Metode Cluster K-Means Untuk Pengelompokkan Fasilitas Sekolah

Faisal Muhammad — 2025-12-14

School Facilities are facilities provided by schools or universities to support activities and can be utilized by students, teachers, students and staff within the scope of a particular education. In order to create good teaching and learning activities (KBM) and support the development process and achievements, good schools or universities must have classroom facilities, laboratories, libraries, canteens, places of worship and fields. By applying data mining and utilizing the data sources obtained and the application of the K-Means cluster method, information related to school facilities can be drawn. The number of clusters obtained is 2 clusters with the number of squares according to the cluster of 76.0%.

Teknik Data Mining Dengan Menggunakan Algoritma Decision Tree Untuk Mengetahui Pola Pemahaman Mahasiswa Pada Matakuliah Pemrograman

Sri Novida Sari — 2025-12-25

Medan State Polytechnic, as one of the leading vocational universities in Medan City, plays a crucial role in producing graduates who are ready to work and possess applied competencies according to industry needs. One of the strategic departments is the Computer Engineering and Informatics Department, which focuses on developing students' abilities in technology and programming. Programming courses are an important foundation in developing students' analytical and logical skills. However, many students still experience difficulties in understanding basic programming concepts, which results in low academic achievement and learning motivation. This study aims to identify patterns of student understanding in programming courses using the Decision Tree algorithm as a classification method. Through a data mining approach, this study attempts to extract hidden patterns from students' academic data to identify factors that influence their level of understanding. The Decision Tree algorithm was chosen because it is able to produce classification models that are easy to understand and interpret, and is effective in handling both categorical and numerical data. The research data was processed using Google Collaboratory with the help of the scikit-learn library. The testing process was carried out through the formation of a classification model, decision tree visualization, and confusion matrix analysis to measure model performance. Based on the test results, an accuracy value of 50% and an F1-score of 51.68% were obtained, indicating that the Decision Tree model has a good ability to predict and classify students' level of understanding of programming courses. Overall, this research provides an important contribution to the development of data-based learning strategies in vocational education environments. Through the results obtained, lecturers are expected to be able to adjust teaching methods according to student characteristics and abilities, so that the learning process becomes more adaptive, effective, and has a positive impact on improving student understanding of programming courses.

Analisis Dan Prediksi Hasil Pertandingan Dota 2 Menggunakan Fuzzy Tsukamoto

Muhammad Arief Adidharma Tan — 2025-12-25

Predicting the outcome of a Dota 2 match is a complex problem because it is influenced by many dynamic variables that change at each stage of the game. This study aims to analyze and predict the probability of winning a Dota 2 match using the Fuzzy Tsukamoto method based on three main variables: Hero Win Rate, Number of Kills, and Tower Destroyed. The fuzzy model was constructed using triangular and trapezoidal membership functions, with variable weights adjusted for the early game, mid game, and late game.

Test results show that in the early game, the Hero Win Rate variable has the most dominant influence on the probability of winning, with a weight of 0.7. In the mid game, the number of kills and tower destruction begin to have a significant impact, while in the late game, towers and kills become the primary determinants of the probability of winning. The proposed system is able to generate different percentages of the probability of winning at each stage of the game and logically reflect the dynamics of the Dota 2 game.

Based on these results, the Fuzzy Tsukamoto method is considered capable of handling uncertainty in Dota 2 match prediction and provides more flexible results than deterministic approaches, although it still depends on the quality of the dataset and the fuzzy rules used.

Penerapan Algoritma K-Means dan Apriori dalam Manajemen Stok UMKM Toko Sembako Berbasis Analisis BCG Matrix

Virdyra Tasril — 2025-12-25

This study aims to analyze purchasing patterns at Toko Sembako HAS in Medan City, Medan Polonia District, using a Hybrid Data Mining approach that combines K-Means and Apriori algorithms. The dataset consists of 75,294 items sold over a 7-month period. The research workflow began with problem identification, literature review, data collection, and pre-processing, followed by algorithm implementation to produce product clustering and association patterns. Data normalization was performed using the Min-Max method to align the scales of Quantity and Profit, ensuring accurate K-Means clustering. The K-Means clustering combined with BCG Matrix categorized products into Stars, Cash Cows, Question Marks, and Dogs. Products such as Indomie and Mie Sedap were classified as Stars with high sales volume and medium-high profitability, while Minyak Curah and Beras were Cash Cows with moderate sales volume but the highest profitability. The Apriori algorithm revealed hidden purchasing patterns, with the highest Lift Ratio of 1.48 observed for the pair Pampers S and Mie Sedap, indicating a strong correlation within the young family segment. The hybrid approach provides strategic insights: K-Means supports inventory management and product segmentation, while Apriori guides marketing strategies such as product bundling and store layout. However, combinations of Cash-Cows and Question Marks yielded Lift Ratios below 1, indicating insignificant associations. The results demonstrate that this hybrid approach enhances understanding of consumer behavior and supports data-driven decisions to optimize sales and profitability.