Discovering Enrollment Knowledge in University Databases

Arun P. Sanjeev and Jan M. Zytkow, Wichita State University

We describe a data mining application in large databases of student records. Our goal has been to discover knowledge useful in understanding the university enrollment and to find ways to increase it. We demonstrate a combination of automated discovery with human involvement in the discovery process. Human operators formulate open questions and interpret the knowledge discovered by the automated discovery system. Some surprising discoveries we have made have led to the repeated cycle of asking questions, running the automated search, and interpreting new results. In this paper we focus on several findings. We show that good high school students are the best source of large numbers of credit hours, but that some of these students drop out, causing significant enrollment losses. We examine the effect of financial aid on retention. We demonstrate that remedial instruction does not seem to help retain the academically under-prepared students. Our results have been surprisingly stable when we used the Fall '87 cohort to verify the findings obtained from the cohort of Fall '86. We have presented our findings to university administrators in a number of meetings. The discovered knowledge can affect decision making and policy formation.

This page is copyrighted by AAAI. All rights reserved. Your use of this site constitutes acceptance of all of AAAI's terms and conditions and privacy policy.