The IRMA Community
Newsletters
Research IRM
Click a keyword to search titles using our InfoSci-OnDemand powered search:
|
Data Mining in Higher Education: Mining Student Data to Predict Academic Persistence
Abstract
Data on pre-undergraduate students for the period Fall 1998 to Spring 2005 was mined to predict persistence for the Fall 2005 Semester. Eight data mining algorithms were used. The results were subjected to a vote in order to arrive at the final prediction. It was expected that of 394 students who spent their first and second semesters (Term1 and Term2) in the undergraduate preparatory year in Fall 2004 and Spring 2005, 50 of them will not persist in Fall 2005. Because the target dataset of 394 records was comparatively small compared to the training and testing dataset of 2279 records (from Fall 1998 to Spring 2004), use of the average risk estimate of 14% and standard error of 1.5% (see section 3.1) is cautioned. This paper presents the background information and the methodology adopted. It also discussed the data mining models used, their performance, and the results obtained. The paper further discusses the results obtained, and concludes with consideration of limitations, management recommendations, and direction for future work.
|
|