A Study on Prediction of Academic Performance based on Current Learning Records of a Language Class using Blended Learning

Abstract

In this paper, we describe a classification method that does not rely on historic data to predict changes in student academic performance, and therefore predict if a student will fail a class or not. By classifying students into groups given their grades, and extracting the common features in between them, it is possible to use those common features to predict if other students that share common characteristics will fall into the same classification groups. As well, those same common features can be used to help students improve their academic performance.