Evaluating Weekly Predictions of At-Risk Students at The Open University: Results and Issues

Improving student retention rates is a critical task not only for traditional universities but particularly in distance learning courses, which are in recent years rapidly gaining in popularity. Early indications of potential student failure enable the tutor to provide the student with appropriate assistance, which might improve the student’s chances of passing the course. Collated results for a course cohort can also assist course teams to identify problem areas in the educational materials and make improvements for future course presentations. Recent work at the Open University (OU) has focused on improving student retention by predicting which students are at risk of failing. In this paper we present the models implemented at the OU, evaluate these models on a selected course and discuss the issues of creating the predictive models based on historical data, particularly mapping the content of the current presentation to the previous one. These models were initially tested on two courses and later extended to ten courses.