Educational data mining: A mining model for developing students' programming skills

Abstract

Educational data mining (EDM) is a branch of data mining and machine learning research to develop new ways to analysis educational data from an educational system. Recently, EDM an rising field of data mining and educational systems. EDM uses traditional mining algorithms to analyses educational data in order to understand and improve the students' learning process. In this paper, we present a decision tree (DT) based mining model for developing students C programming skills. DT is a rule set, which is a top-down recursive divide and conquer mining algorithm in supervised learning. We have collected data from 70 students of Structured Programming Language (SLP) course and generated two datasets StuBehEduInfo and QuickTestInfo. The StuBehEduInfo dataset contains student behavioural and past educational attributes. The QuickTestInfo dataset contains simple C programming questions. Then using these datasets we built two decision trees that can classify the students into three groups (Good, Average, & poor), so we can take extra care of the weakest students. The proposed decision tree models can correctly classify 87% students.