Jurnal Sains dan Teknologi Informasi

Pengembangan Aplikasi Mobile Muhasabah Untuk Evaluasi Diri Harian Berbasis Kotlin Android Dengan Pendekatan Self-Assessment

Raafi Hilmi — 2026-03-31

Self-reflection (muhasabah) is crucial for fostering self-awareness and improving one’s quality of life, yet it has not become a daily habit—particularly among younger generations. Limited awareness, time constraints, and the lack of practical digital tools are key obstacles. This study aims to design an Android-based “Muhasabah Harian” application that enables users to conduct independent self-evaluation through a self-assessment approach. Development follows the Waterfall model, covering requirement analysis, design, implementation, testing, and support. Data were gathered through literature review, observation of similar applications, and interviews with teachers and students at SDN Panunggangan 2 Pinang. The system adopts the MVVM architecture, Jetpack Compose, and Kotlin. The application offers two modes: guest (offline) with local storage via Room Database and registered user (online) with storage in Firebase Firestore. Key features include reflection entry, evaluation history, progress graphs, and reminder notifications. The resulting application is expected to provide a practical and effective medium for daily self-reflection.

Evaluasi Kepuasan Pengguna Sistem Informasi Akademik Menggunakan Metode End User Computing Satisfaction

Prama Dita Laura — 2026-03-31

Digital transformation in higher education requires academic data management that is not only technically functional but also capable of satisfying its users. This study aims to evaluate user satisfaction with the Academic Information System (SIAKAD) at Mitra Gama Institute of Technology using the End User Computing Satisfaction (EUCS) framework. The research method used was descriptive quantitative, involving 28 active user respondents selected through purposive sampling. The evaluation was conducted across five main dimensions: Content, Accuracy, Format, Ease of Use, and Timeliness. Data were collected using a Likert-scale questionnaire that had been validated using Pearson’s correlation and tested for reliability with Cronbach’s Alpha (0.947). The analysis results showed that user satisfaction levels varied across each dimension. The Content dimension received the highest score of 71.42%, followed by the Accuracy dimension at 60.72%, both of which fell into the “good” category. However, the Format dimension recorded the lowest score of 53.55%, rated as “fair,” indicating significant complaints regarding the user interface, which was deemed unresponsive and rigid. Meanwhile, the Timeliness dimension scored 43.74% and Ease of Use scored 41.97%. The study’s findings concluded that while SIAKAD has functionally been able to provide relevant content, there is an urgent need to improve visual aesthetics, navigation flow efficiency, and server performance stability—especially during peak access periods. The proposed strategic recommendations include migrating to a more modern interface design and optimizing real-time data synchronization to holistically enhance the user experience.

Klasifikasi Penyakit Mata Menggunakan Random Forest Dengan Optimasi Hyperparameter RandomSearchCV

Muh. Fatkhi Alexander — 2026-03-31

Eye diseases such as cataracts, diabetic retinopathy and glaucoma are the leading causes of visual impairment and blindness worldwide, so early detection through medical image analysis is essential to prevent complications and permanent vision loss. The development of artificial intelligence and machine learning technology provides great opportunities to help medical personnel carry out diagnoses more quickly, accurately and efficiently. This research aims to develop an eye disease classification model using the Random Forest algorithm with hyperparameter optimization to differentiate four eye conditions, namely cataract, diabetic retinopathy, glaucoma, and normal. The dataset used is sourced from the public and consists of eye fundus images that have gone through preprocessing and feature extraction to improve data quality. The data was divided into training and testing, then the Random Forest model was trained with hyperparameter optimization using RandomizedSearchCV for 20 iterations and 5-fold cross-validation to obtain the best parameter combination. The best model achieved an accuracy of 80.92% on testing data with a macro ROC-AUC value of 0.9422, where the best performance was obtained in the classification of diabetic retinopathy with a precision of 99.54%, recall of 99.09%, and ROC-AUC of 1.0000. In addition, macro specificity reached 93.66%, indicating the model's good ability to identify negative cases correctly. The research results show that the Random Forest approach with hyperparameter optimization has excellent performance for eye disease classification and has the potential to be implemented as an artificial intelligence-based medical diagnosis support system in health care facilities.

Optimalisasi Rasio Data pada K-Nearest Neighbor untuk Klasifikasi Multikelas Tingkat Obesitas Populasi Dewasa

Dini Aprilia Langnegara — 2026-03-31

Obesity is a complex health issue that needs a strategy for assessing its severity to facilitate earlier recognition. One can determine an individual's obesity classification by analyzing their dietary habits, level of physical activity, and overall health status. This research aims to ascertain the K-Nearest Neighbor (KNN) algorithm's efficacy in accurately classifying seven various phases of obesity. The dataset employed for predicting obesity consisted of 2,111 samples drawn from a population of both genders. For KNN testing, the dataset was divided into training and test data, with the test data allocated over three separate scenarios, including varying ratios. The ratios of 70:30, 80:20, and 90:10 were utilized in these circumstances, respectively. The value of k was varied from k=2 to k=10. The optimal configuration was achieved with a 90:10 data split ratio and a k value of 2, as evidenced by the test results. This setup concurrently attained an accuracy of 90.05%, a precision of 90.56%, a recall of 89.80%, and an F1 score of 90.18%. This categorization error was most prominent when comparing the Normal Weight category to the Class I Overweight group. A properly preprocessed KNN algorithm can attain competitive accuracy over 90 percent in classifying population obesity levels, as demonstrated by this study's findings.

Pengembangan Sistem Informasi Manajemen Bimbingan Belajar Berbasis Web dengan Pendekatan Data Mining untuk Analisis Performa Siswa Menggunakan Framework Laravel

Atika Fauziyyah — 2026-03-31

This research is motivated by the need for a system capable of managing tutoring institution data effectively while accurately analyzing student performance. The main problem at Brigade Nusantara is that data management is still conducted manually and there is no data-based analysis available to support decision-making processes. This study aims to develop a web-based Tutoring Management Information System using the Laravel framework with a data mining approach to analyze student performance. The system development method includes the stages of requirements analysis, system design, implementation, and testing. The data mining approach is utilized to process students’ historical data in order to generate information in the form of performance patterns, learning progress levels, and recommendations for academic improvement. The results of the study indicate that the developed system is capable of improving data management efficiency, facilitating the monitoring of student progress, and supporting accurate and data-driven decision making. Therefore, this system is expected to enhance the overall quality of tutoring services and assist administrators in developing more effective and targeted learning strategies.