Influence of Feature Selection on the Prediction of Student Performance

Abstract

In the process of students evaluation most academic institutions assume that the performance of student is the major criteria. Machine learning offers different techniques used in several fields of education including student performance. This paper presents analytic study to find the most affective attributes related to student academic performance by applying classification algorithms on a collected student`s data. The data collected from Computer science department in Tikrit University-Iraq. The attributes labeled into four categories (personal, family, study, and online activities) then a combination of classification models tested on each type of the attributes. This study aims to give the academic educators good understanding of the obstacles facing their student and could affect their grades. The subset of “study-attributes” resulted best accuracy in all models