Applying predictive analytics in identifying students at risk: A case study

Abstract

In this article, a case study is presented of an institutional modelling project whereby the most appropriate learning algorithm for the prediction of students dropping out before or in the second year of study was identified and deployed. This second-year dropout model was applied at programme level using pre-university information and first semester data derived from the Higher Education Data Analyzer (HEDA 1) management information reporting and decision support environment at the Cape Peninsula University of Technology. An open source platform, namely Konstanz Information Miner (KNIME 2), was used to perform the predictive modelling. The results from the model were used in HEDA automatically to recognize students with a high probability of dropping out by the second year of study. Being able to identify such students will enable universities, and in particular programme owners, to implement targeted intervention strategies to assist the students at risk and improve success rates.