OPTIMASI ALGORITMA C4.5 DAN NAIVE BAYES MENGGUNAKAN K-MEANS UNTUK PREDIKSI KELULUSAN MAHASISWA

Abstract

Today's education system demands quality-oriented education. The quality of Indonesian higher education is measured based on accreditation issued by the Badan Akreditasi Nasional Perguruan Tinggi. One indicator of success in the process of managing education on higher education is the period of student graduation. Undergraduate students have a study load of 144 credits which can be taken in 8 semesters. but in fact, there are still many students who cannot complete their studies for 8 semesters due to various factors such as lack of motivation, intelligence factors, and economic factors. There is a need for continuous monitoring and evaluation of periods in student graduation using the C4.5 and Naive Bayes algorithms. Optimization is needed to increase the accuracy value of the C4.5 and Naive Bayes algorithms by using K-means for the data discretization process. The experimental result show C4.5 algorithm with K-means produces an accuracy value of 89.74%, a precision value of 90.60%, and a recall value of 98.00% while Naive Bayes with K-means produces an accuracy value of 80.73%, a precision value of 89.60%, a value recall of 87.20%. The comparison of two classification algorithms combined with K-means shows that the C4.5 algorithm has a better performance than Naive Bayes.