Abstract:

Students on introductory programming courses find the abstract programming concepts difficult to understand, and often lack the intrinsic motivation. Hence, the dropout rates are usually quite high. The teachers have difficulties in identifying the students that are at risk of dropping out – in fact, the identification of those students typically cannot be done before the end of the course. The identification is especially difficult on large courses where the teachers do not know the individual students. Still, it is reasonable to assume that the students' study habits and general performance stay relatively similar throughout the course. In this paper, we show that by utilizing novel methods of learning analytics together with a learning management system (LMS), it is possible to identify the students-at-risk as early as after the first two or three weeks of the course. Commonly, all of the course events, such as exercise submissions, lecture attendances and weekly assignments, are recorded by the LMS. The analysis utilizes this automatically collected data to identify the students-at-risk. We describe the method used for analysis in detail, and show that the student performance data in the course can be used to reliably predict those who are at risk already at early stages of the course.