SISTEM KLASIFIKASI PREDIKSI KELULUSAN MAHASISWA MENGGUNAKAN ALGORITMA C4.5 DENGAN PERBANDINGAN RAPIDMINER DAN WEKA

Abstract

The student graduation prediction system is an important indicator in evaluating academic success in higher education. This study compares the performance of the C4.5 algorithm using two data mining software, namely RapidMiner and Weka , with data in the form of Semester Achievement Index (IPS) and the number of Semester Credit Units (SKS) from semesters I to V. Non-numeric data is removed because it is not relevant in the classification process. The process in RapidMiner includes the Read Excel, Set Role, Decision Tree, Apply Model, and Performance stages. While in Weka , the process is carried out through data conversion to ARFF format, attribute selection, application of the J48 algorithm, and testing using the Supplied Test Set. The test results show that both software produce the same accuracy, namely 86.67%, with 26 data correctly classified out of 30 test data. Despite identical accuracy, RapidMiner is considered superior in terms of visualization and ease of modeling, while Weka is lighter and more efficient on simple numerical data. This system is expected to be a reference in supporting academic evaluation and helping predict student graduation more effectively and on time.