The Analysis of Student Performance Using Data Mining

Abstract

This paper presents the study of data mining in the education industry to model the performance for students enrolled in university. Two algorithms of data mining were used. Firstly, a descriptive task based on the K-means algorithm was utilized to select several student clusters. Secondly, a classification task supported two classification techniques, known as decision tree and Naïve Bayes, to predict the dropout because of poor performance in a student’s first four semesters. The student academic data collected during the admission process of those students were used to train and test the models, which were assessed using a cross-validation technique. Experimental results show that the prediction of drop out student is improved, and student performance is monitored when the data from the previous academic enrollment are added.