Using Data Mining Techniques to Assess Students’ Answer Predictions

Abstract

Estimating students´ knowledge and performance, modeling their learning behaviors, and discovering and analyzing their different characteristics are some of the main tasks in the field of research called educational data mining (EDM). According to Chounta (2017), the predicted probabilities that a student will answer a question correctly can provide some insights into the student´s knowledge. Based on this point of departure, the main objective of this paper is to apply different data mining techniques to predict the probabilities that students will answer questions correctly by using their interaction records with a web-based learning platform called Hypocampus. Five different machine learning algorithms and a rich context model were used on the Hypocampus dataset. The results of our evaluation indicate that the gradient-boosted tree and the XGboost algorithms are best in predicting the correctness of the student’s answer.