Penerapan Algoritma Decision Tree untuk Memprediksi Prestasi Belajar Siswa di SMP Negeri 1 Susukan

Abstract

Student academic achievement is an important indicator for evaluating the effectiveness of the learning process. The application of data mining techniques can help schools identify students' academic performance through classification models. This study aims to implement the Decision Tree (J48) algorithm to predict student academic achievement at SMP Negeri 1 Susukan and to compare the performance of the model using several Percentage Split scenarios. The dataset consisted of academic records from 263 students, including seven predictor attributes: Assignment Score, Midterm Examination Score, Final Examination Score, Attendance Rate, Daily Social Media Usage, Behavior, and Report Card Score, with Academic Achievement as the target class. The classification process was carried out using WEKA 3.8 with three Percentage Split scenarios: 80%:20%, 85%:15%, and 90%:10%. Model performance was evaluated using Accuracy, Precision, Recall, F1-Score, and Kappa Statistic. The experimental results showed that all scenarios achieved an accuracy of more than 96%, with the 80% training and 20% testing scenario producing the best performance, achieving an accuracy of 98.11%, precision of 0.982, recall of 0.981, and F1-score of 0.981. The resulting decision tree identified Report Card Score as the most influential attribute in the classification process. These findings demonstrate that the Decision Tree (J48) algorithm is capable of producing an accurate and interpretable classification model, making it a promising decision-support tool for identifying students' academic achievement and supporting educational decision-making in schools.