Predicting First-Year Computer Science Students Drop-Out with Machine Learning Methods: A Case Study

In this paper, we describe the results of the educational machine learning case study with the aim to predict first-year computer science students' dropout in the Virumaa College of Tallinn University of Technology and determine factors that influence dropout rates. In this study two different datasets are used: (1) data obtained from the TalTech study information system; (2) students’ history and study results collected in Virumaa College. To build predictive models, the following machine learning algorithms are applied: Naive Bayes, decision trees, Logistic Regression, Support Vector Machines and Neural Networks. As a result of this study were evaluated how the dropout prediction accuracies change from the moment of the students’ admission to the end of the first semester. We found, that data that were available about students before enrollment allowed to predict dropout with 70% of accuracy. Using data that obtained from first semester allowed to rise prediction accuracy to 90%. Besides, the factors were determined that are related with drop-out and that are not. Any higher education institution can conduct a similar study, since it is conducted on publicly available data from the official academic information environment.