Bulletin of Information Technology (BIT)

Klasifikasi Genre Musik Menggunakan Machine Learning

Garda Zidane Dhamara — 2025-09-26

This study examines the implementation of music genre classification using Machine Learning to develop an accurate and efficient music recommendation application. The main problem addressed is the automatic identification of music genres to improve recommendation personalization. The method used involves applying Machine Learning algorithms to a music dataset. The objective of this research is to build a system capable of automatically classifying music genres and serving as a foundation for a smarter recommendation system. Preliminary results indicate that Machine Learning is effective in music grouping, which will contribute to increased recommendation accuracy. This research is expected to make a significant contribution to the development of intelligent music applications.

Implementasi Naïve Bayes untuk Klasifikasi Peminatan Program Studi pada Penerimaan Mahasiswa Baru di Fakultas Ilmu Komputer Unika

Munawirah Munawirah — 2025-09-26

Fakultas Ilmu Komputer Universitas Tomakaka memiliki dua program studi, yaitu Sistem Informasi dan Teknik Informatika. Namun, dalam praktiknya, calon mahasiswa baru sering mengalami kebingungan dalam menentukan jurusan yang sesuai dengan kemampuan dan latar belakang akademiknya. Pemilihan program studi umumnya didasarkan pada tren jurusan favorit, dorongan eksternal, atau preferensi sosial tanpa mempertimbangkan jurusan asal di sekolah sebelumnya. Kondisi tersebut berpotensi menimbulkan ketidaksesuaian minat yang berdampak pada risiko penurunan motivasi belajar, pindah jurusan, berhenti kuliah, atau mengalami hambatan selama masa studi. Penelitian ini bertujuan untuk mengembangkan sistem rekomendasi program studi menggunakan metode klasifikasi Naïve Bayes guna memprediksi kecenderungan peminatan program studi berdasarkan atribut input seperti jenis kelamin, asal sekolah, dan jurusan asal sekolah. Dataset yang digunakan merupakan data historis penerimaan mahasiswa baru Fakultas Ilmu Komputer Universitas Tomakaka sejak tahun akademik 2015/2016 hingga 2024/2025, sebanyak 1.046 entri data. Proses analisis mencakup tahapan data mining, mulai dari seleksi dan pembersihan data, pembagian data latih dan data uji (80:20), hingga evaluasi performa menggunakan metode Confusion Matrix. Hasil evaluasi menunjukkan akurasi sebesar 87,14%, presisi 89,91%, recall 87,70%, dan F1-score 88,76%. Model ini diimplementasikan ke dalam aplikasi berbasis website menggunakan framework Flask, guna mempermudah pemberian rekomendasi jurusan secara real-time. Pendekatan ini memberikan kontribusi sistem rekomendasi berbasis data yang membantu institusi dalam memetakan minat mahasiswa, menyusun strategi promosi yang tepat sasaran, serta memberikan intervensi awal terhadap pilihan program studi mahasiswa baru yang kurang sesuai.

Face Recognition Motorcycle Rider Registration System for Rider Data Management

Insan Taufik — 2025-09-26

This research aims to develop a motorcycle rider registration system using facial recognition technology that can improve the efficiency of rider data management. This system is designed to identify and authenticate riders with high accuracy, thereby simplifying the registration and monitoring process. The methods used in this research include collecting rider facial data through cameras, image processing for feature extraction, and implementing a facial recognition algorithm. Testing was conducted in several locations with varying lighting conditions and viewing angles to ensure the system's robustness. The results show that the developed system is capable of achieving facial recognition accuracy of up to 95%. In addition, this system provides an intuitive user interface to facilitate the registration and data management process. With the implementation of this system, it is expected to reduce the time and costs required in managing motorcycle rider data, as well as improve safety and comfort while riding.

Rancang Bangun Prototype Sistem Monitoring Dan Kontrol Tanaman Hidroponik Berbasis Internet of Things (IoT) Menggunakan Microcontroller ESP32

Veri Ferdiansyah — 2025-09-26

Hydroponic farming has emerged as an innovative solution to address land limitations and support sustainable food production. However, its success highly depends on consistent monitoring of light intensity and nutrient water availability. This study aims to design and develop a prototype monitoring and control system for hydroponic plants based on the Internet of Things (IoT) using the ESP32 microcontroller. The system employs an ultrasonic sensor to measure water level, a Light Dependent Resistor (LDR) sensor to detect light intensity, a 12V DC water pump, and LED grow lights as actuators. Environmental condition data is transmitted in real-time to the Blynk mobile application, which also provides automatic notifications when anomalies occur, such as low water levels or light intensity falling below the threshold. The development method used is Research and Development (R&D) with the ADDIE model, covering analysis, design, development, implementation, and evaluation stages. Testing results show that the system operates automatically and in real-time, achieving 100% detection accuracy for water level measurements and 98% for light intensity measurements. The implementation of this prototype is expected to improve the efficiency and effectiveness of small-scale hydroponic cultivation and serve as an affordable solution for farmers and the general public to adopt smart farming technology.

Pengembangan dan Implementasi Sistem Deteksi Serangan DDoS Berbasis Algoritma Random Forest

Dedy Kiswanto — 2025-09-26

Serangan Distributed Denial of Service (DDoS) merupakan ancaman serius bagi keamanan jaringan, sementara metode deteksi tradisional seperti threshold-based detection dan signature-based detection memiliki keterbatasan dalam mengenali pola serangan baru maupun anomali lalu lintas yang kompleks. Penelitian ini bertujuan merancang dan mengimplementasikan model prediksi serangan DDoS berbasis algoritma Random Forest yang mampu membedakan trafik normal dan berindikasi serangan secara akurat. Pendekatan Research and Development (R&D) digunakan, meliputi studi literatur, perancangan model, implementasi, serta evaluasi performa menggunakan metrik akurasi, precision, recall, F1-score, confusion matrix, dan learning curve. Berdasarkan hasil evaluasi, model Random Forest menunjukkan kinerja sangat baik dengan akurasi 0,99942 (99,942%). Precision untuk kelas 0 dan 1 masing-masing sebesar 0,99979 dan 0,99884, sedangkan recall mencapai 0,99928 untuk kelas 0 dan 0,99966 untuk kelas 1. Nilai F1-score tinggi, yaitu 0,99953 untuk kelas 0 dan 0,99925 untuk kelas 1, dengan macro average F1-score sebesar 0,99939 dan weighted average sebesar 0,99942, menunjukkan keseimbangan performa pada kedua kelas. Confusion Matrix menunjukkan kesalahan klasifikasi rendah (44 false positive dan 13 false negative dari 99.066 sampel). Analisis learning curve mengungkapkan akurasi pelatihan stabil di atas 0,998, sedangkan akurasi validasi meningkat dari 0,986 pada 10.000 data hingga di atas 0,998 pada 80.000 data, dengan jarak antarkurva semakin kecil. Pola ini menandakan model mampu memanfaatkan data tambahan untuk meningkatkan generalisasi tanpa gejala overfitting atau underfitting. Temuan ini membuktikan bahwa model Random Forest yang dirancang dapat menjadi solusi deteksi dini serangan DDoS yang andal, adaptif, dan berpotensi diintegrasikan dalam sistem keamanan jaringan secara real-time.

Sistem Pemantauan Tanaman Dalam Pot Indoor Dengan Internet of Things

Muhammad Iqbal Setiawan — 2025-09-26

This study discusses the design and implementation of an Internet of Things (IoT)-based indoor potted plant monitoring system, which aims to help users care for plants in a more effective and efficient manner. The system uses an ESP32 microcontroller connected to a DHT22 sensor to measure air temperature and humidity, soil moisture, an LDR to measure light intensity, and a TDS sensor to monitor nutrient levels in the water. Data collected from the sensors is transmitted directly via a WiFi connection to an MQTT broker, displayed on a Node-RED dashboard, and stored in Firebase for historical documentation purposes. This system has two operational modes, manual and automatic, allowing users to control the water pump and grow light directly or let the system operate based on pre-set parameters. Test results show that all sensors function accurately and respond to changes in environmental conditions, thereby improving efficiency in watering and lighting. The advantage of this system lies in the integration of four monitoring parameters into a single platform that is easy to use, flexible, and widely accessible. This research is expected to provide practical solutions for urban agriculture and the development of smart farming at the household level, although further testing on various plant types and environmental conditions is still needed for further refinement

Metode Maut dan Waspas Menentukan Mahasiswa Berprestasi di Universitas Bhayangkara Jakarta Raya dengan Pembobotan ROC

Bernadus Gunawan Sudarsono — 2025-09-26

Becoming an outstanding student in higher education is a positive and proud achievement, reflecting the national education goal of developing students' potential to become educated, creative, and democratic and responsible citizens. Determining outstanding students faces obstacles when prospective candidates excel in some criteria but do not meet the standards in other criteria. To help the evaluation team, an effective decision support system is needed. The Multi-Attribute Utility Theory (MAUT) method with ROC weighting was used to convert various interests into numerical values on a scale of 0-1, with the results showing that student Erwin Sulistiono (A4) had the highest utility value, namely 0.8975. For comparison, the Weighted Aggregated Sum Product Assessment (WASPAS) method was also applied, combining the Weighted Sum Model (WSM) and the Weighted Product Model (WPM), which gave consistent results with MAUT, showing that both methods provide an objective approach in determining outstanding students, although WASPAS with ROC weighting offers higher accuracy by combining the advantages of two scoring approaches.

Sistem Pakar Berbasis AI dengan Artificial Neural Networks untuk Identifikasi Hama & Penyakit Jamur Tiram

Nursuci Putri Husain — 2025-09-26

Oyster mushroom cultivation is an agricultural sector with high economic potential, but its productivity is often disrupted by pests and diseases. Inappropriate management due to farmers' limited knowledge can cause significant losses. This study aims to develop an expert system for oyster mushroom pest and disease diagnosis based on Artificial Neural Networks (ANN), to assist in early identification of emerging disorders. The dataset consists of 150 samples covering a combination of symptoms and disease labels, collected from two different cultivation locations. There are several stages in this study, namely the preprocessing process that includes label encoding, feature normalization using Z-score, and data division in a ratio of 80% for training and 20% for testing. The ANN model was designed using a Multi-Layer Perceptron (MLP) with two hidden layers containing 10 neurons each, a ReLU activation function, an Adam solver, and a maximum iteration of 1000. The test results showed the model has an accuracy rate of 97%, with perfect precision and recall values for most disease classes. This study shows that the ANN approach is able to effectively recognize oyster mushroom disease symptom patterns. This system can be an efficient and adaptive diagnostic tool, and has the potential to be further developed as a smart agricultural technology solution

Sentiment Analysis of User Reviews of Kitalulus Job Search App on Google Play Store Using Machine Learning

Astrid Ayuzi Putri Hendri Hariadi — 2025-09-27

This study seeks to assess the sentiment of user reviews for the "KitaLulus" job search app found on the Google Play Store, utilizing Machine Learning techniques. Given the intensifying competition within the job market, this application serves as a crucial resource for job seekers in Indonesia. The study employs a sentiment analysis method to categorize user reviews into three groups: positive, negative, and neutral. The dataset comprises 20,000 reviews in Indonesian gathered from the Google Play Store. The methodologies used in this study include K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Artificial Neural Network (ANN), Logistic Regression, and Naïve Bayes. The findings indicate that various algorithms demonstrate different levels of accuracy in sentiment classification. It is anticipated that the outcomes of this analysis will offer valuable insights to developers about the quality and effectiveness of the "KitaLulus" application, while also assisting users in making informed decisions prior to utilizing the app. Additionally, this research contributes to the domain of sentiment analysis, particularly concerning job search applications in Indonesia.

Komparasi Model LSTM dan CNN-LSTM untuk Peramalan Curah Hujan di Kota Tangerang Selatan

Uliyatunisa — 2025-09-27

This study compares the performance of Long Short-Term Memory (LSTM) and Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) models for daily rainfall forecasting in South Tangerang City using meteorological data from January 2005 to July 2025. Data from official meteorological stations was processed with mean imputation for missing values and MinMaxScaler normalization. Models were evaluated based on Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Squared Error (MSE), and coefficient of determination R². Results show CNN-LSTM outperforms with RMSE 0.79, MAE 0.63, MSE 0.62, and R² 0.61, compared to LSTM (RMSE 0.83, MAE 0.60, MSE 0.68, R² 0.58). Prediction visualizations confirm CNN-LSTM's accuracy in capturing extreme patterns, with statistically significant differences via t-test. The novelty lies in using a long-term (20-year) dataset for tropical Indonesia, demonstrating the hybrid model's efficacy for complex spatio-temporal predictions. Findings support flood early warning systems and water resource management, recommending additional climate variable integration for further development.

Penerapan Algoritma Bellman-Ford Untuk Optimisasi Pengendara Dalam Menentukan Rute Terpendek UMKM Di Kabupaten Padang Lawas

Putri Indah Julia Hasibuan — 2025-09-27

This study designs a web-based Geographic Information System (GIS) using the Bellman-Ford algorithm to determine the shortest route for Micro, Small, and Medium Enterprises (MSMEs) in Padang Lawas Regency. The main problem faced by MSMEs is the limited and insufficient use of information technology. This condition makes it difficult for the public to find strategic MSME locations and the fastest routes to business sites, leading to low competitiveness and marketing inefficiency. To address this issue, the system was developed using the waterfall model and integrated with Leaflet JS technology, enabling broad accessibility through the web without additional installation. The Bellman-Ford algorithm was chosen for its ability to calculate the shortest path even when negative weights are present in the graph. Test results show that the optimal route obtained is 1.795 km, more efficient compared to an alternative route of 2.563 km, providing a distance saving of about 30%. The system has proven capable of delivering fast and accurate route recommendations while simultaneously presenting MSME location information interactively. The novelty of this research lies in the integration of Bellman-Ford with interactive web-based digital maps specifically for MSME promotion, which has rarely been applied in regional contexts. The purpose of this study is to improve marketing efficiency, expand accessibility, and strengthen the competitiveness of MSMEs in Padang Lawas. Furthermore, this research is expected to make a real contribution to the community in finding MSMEs more quickly and accurately..

Sistem Informasi Geografis Pemetaan Daerah Rawan Pangan Pada Dinas Ketahanan Pangan di Kabupaten Labuhanbatu Utara Dengan Algoritma Dijkstra Berbasis Web

Rico Almanda — 2025-09-27

Food security is a strategic aspect that determines community welfare and regional stability. In North Labuhanbatu Regency, 24 out of 82 villages and 8 sub-districts are still categorized as food-insecure. The main challenges include manual mapping processes and difficulties in determining the nearest food distribution routes, which make government interventions less efficient. This study aims to develop a web-based Geographic Information System (GIS) to map food-insecure areas and calculate the fastest distribution routes using Dijkstra’s algorithm. The system was developed using the waterfall model through observation, interviews, and literature studies. The test results show that the system can interactively visualize 24 food-insecure villages and recommend distribution routes with an average distance of 37.04 km and an average travel time of 37.35 minutes. These results are more efficient than manual methods, which tend to generate longer routes and higher travel times. The main contribution of this research is the application of Dijkstra’s algorithm in a web-based GIS for mapping food-insecure villages, which has not been implemented previously in North Labuhanbatu Regency. This finding is expected to support local governments in making strategic decisions regarding the acceleration of food distribution, thereby improving public services and strengthening community food security.