Online Tests and Predictive Analytics

Abstract

We present a framework that estimates students’ exam performance in the context of mathematics lectures. For the prediction, we apply a multi-variate nonlinear regression implemented in MATLAB based on the results of five tests acting as independent data. The electronic tests were implemented by means of online assessment system STACK. The tests cover all major topics and are evenly distributed throughout the lecture period. The performance of the approach is quantitatively tested on student groups. In the future, the insights gained in this way will serve as a starting point for prespective analytics.