Educational data mining to evaluate student evasion in information technology courses

Abstract

The evasion of students at the higher level is a serious problem. This work introduces a method based on graph-oriented Database to understand and to perform identification of students with high evasion risk. Our approach uses exclusively the students' academic history, something easy to get to apply the proposed method. It was calculated the similarity between current students and evaded students from previous classes. The results identified patterns and showed that is possible to accurately identify 73\% the final situation of the student. The proposed method may indicate the taking of individual actions directed at students at risk, as well as the planning of future actions, to reduce the amount of evaded students.