Predicting student dropout: A machine learning approach

Abstract

We perform two approaches of machine learning, logistic regressions and decision trees, to predict student dropout at the Karlsruhe Institute of Technology (KIT). The models are computed on the basis of examination data, i.e. data available at all universities without the need of specific collection. Therefore, we propose a methodical approach that may be put in practice with relative ease at other institutions. We find decision trees to produce slightly better results than logistic regressions. However, both methods yield high prediction accuracies of up to 95% after three semesters. A classification with more than 83% accuracy is already possible after the first semester.

Topics & Concepts