Prediction of Students’ Academic Performance based on Tracking logs

Abstract

In this paper, we predict students’ academic performance based on tracking log of students’ learning activities. We compare the prediction of six datasets from Kyoto University (KU), National Central University (NCU), and Chung Yuan Christian University (CYCU) by eight classification models. We use the evaluators of accuracy, recall, precision, F1-score, and Area Under the Curve (AUC) of Receiver Operating Characteristic (ROC). According to the prediction results, we found that sample size and feature category influence the prediction performance of classification. We also found that the significant features based on Pearson correlation analysis have greatly influence on the prediction performance of classification.