Data mining as a tool for generating growth in a shrinking university market

In this paper data mining methods are used in the context of generating growth in shrinking university markets. Growth in this context is associated with a reduction in course withdrawal rates, increases in the transfer of undergraduate students to postgraduate study and the timely completion of students. This study uses a data base of nearly 55,000 student records, allowing the use of decision trees to better understand the factors that lead to withdrawal. In most cases it is found that students who perform well are unlikely to withdraw from their courses, but there is some evidence to suggest that this is not always the case. A self-organising map is used to identify clusters of students who are more likely to withdraw from their courses. The results suggest that although on average performance mediates the effect of student and enrolment characteristics on course withdrawal, there are exceptions to this rule for certain important student groups. This information can be used to grow what is likely to become a shrinking university market in 2013