Performance Comparison of Data Mining Classification Algorithms for Early Warning System of Students Graduation Timeliness

Abstract

Observation of growing academic data can be carried using data mining methods, for example, to obtain knowledge related to the determinants of timeliness of students graduation. This study conducted a performance comparison of the classification algorithms using decision tree (DT), support vector machine (SVM), and artificial neural network (ANN). This study used students academic data from Faculty of Engineering, Universitas Jenderal Soedirman in the 2014/2015 odd semester until the 2017/2018 odd semester and the attributes that conform to the academic regulations. The analytical method used is CRISP-DM. The results showed that SVM provided the best performance in an accuracy of 90.55% and AUC of 0.959, compared to other algorithms. A Model with SVM algorithm can be implemented in an early warning system for timeliness of student graduation.

Keywords

• graduation timeliness
• data mining
• data classification
• data mining algorithms comparison