Data mining techniques in higher education research : The example of student retention.

Data Mining has been used for more than a decade in a variety of differing environments. It takes an inductive approach to data analysis in that it is concerned with the extraction of patterns from the data often without preconceived ideas. Data mining is part of the field of Business Intelligence, a subject area that the author is familiar with and has taught for many years. He believes that the application of data mining techniques has much to offer within the context of higher education. However, there is little evidence that these well established techniques have previously been applied to the sphere of higher education. Student retention is a hot issue in higher education at the moment. It is for this reason that the author chose to establish the power of data mining techniques in higher education using the examination of student retention issues as a vehicle. The field of student retention has been well documented over the years. Contemporary authors such as McGivney (1996), Moxley et al (2001), Yorke (1999) and Yorke & Longden (2004) have examined strategies and derived intervention techniques aimed at assisting students to adapt to university life. As the proportion of students entering Higher Education has increased there has been an increasing awareness that universities need to adapt to the changing profile of these students. The data was collected via an online questionnaire administered to a large group of computing students at Sheffield Hallam University and similar institutions. The collected data was explored using Data Mining techniques including Decision Trees, Market Basket Analysis and Cluster Analysis.This study sought to explore interrelationships between factors that contribute to student attrition and hence establish the demographics of at-risk students. The use of data mining techniques was found to be highly effective, having found most of the primary issues established in previous research. It went on to find the strongest relationships between them, corresponding well to findings from previous research using standard statistical techniques. The author believes that he has established the power of data mining techniques in higher education and recommends further areas where it could be used profitably.