Optimasi Algoritma Decision Tree Dengan Seleksi Fitur Dalam Klasifikasi Prestasi Akademik Siswa Sekolah

Abstract

Quality education is an important foundation for the progress of the nation. In an effort to improve the quality of education, student academic achievement is a significant indicator. With the development of technology, the application of machine learning algorithms, such as Decision Tree, allows for more accurate prediction of student academic achievement. This study aims to optimize the Decision Tree C4.5 algorithm through a combined feature selection between Gain Ratio and Chi-Square to improve the performance of student academic achievement classification. The research data were collected from students of SMA N 1 Blahbatuh and went through a preprocessing process, feature selection, and evaluation using accuracy, precision, recall, and F1-Score metrics. The results showed that the combined feature selection method succeeded in improving the performance of the C4.5 algorithm with an accuracy of 82.2%, much higher than the model without feature selection (54.2%). The implementation of a web-based system was also developed to support practical predictions. Thus, the results of this study contribute to the development of educational data analysis methods to improve the quality of education in the future.