Volume 8

Volume 8, Number 19

April 29, 2010

thumb
7 pages358 K bytes

Myth Busting: Using Data Mining to Refute Link between Transfer Students and Retention Risk


Brenda McAleer
The University of Maine
Augusta, ME 04330 USA

Joseph S. Szakas
The University of Maine
Augusta, ME 04330 USA

Abstract: In the past few years, universities have become much more involved in outcomes assessment. Outside of the classroom analysis of learning outcomes, an investigation is performed into the use of current data mining tools to assess the issue of student retention within the Computer Information Systems (CIS) department. Utilizing both a historical dataset of CIS students over a 10 year period, and a current student dataset, this analysis specifically deals with the following questions: 1. How can we use the past to predict retention risk of the future students? 2. Do students who transfer CIS courses (core or elective) have an increased retention risk? The data mining tool was the Oracle Data MiningTM Package used to perform tasks as classification (Naïve Bayesian and support vector machine), and attribute importance.

Keywords: data mining, attribute importance, naïve Bayesian model, support vector machine model, predicting retention risk, transfer students, assessment

Download this issue:   ISEDJ.8(19).McAleer.pdf   (Adobe PDF, 7 pages, 358 K bytes)

Preview the contents:   McAleer.j.txt   (ASCII txt, 17 K bytes)


Recommended Citation: McAleer and Szakas (2010). Myth Busting: Using Data Mining to Refute Link between Transfer Students and Retention Risk. Information Systems Education Journal, 8 (19). http://isedj.org/8/19/. ISSN: 1545-679X. (A preliminary version appears in The Proceedings of ISECON 2008: §3154. ISSN: 1542-7382.)