Welcome to the InfoSci Platform
Reference Hub16
A Predictive Analytics Approach to Building a Decision Support System for Improving Graduation Rates at a Four-Year College

A Predictive Analytics Approach to Building a Decision Support System for Improving Graduation Rates at a Four-Year College

Xuan Wang, Helmut Schneider, Kenneth R. Walsh
Copyright: © 2020 |Volume: 32 |Issue: 4 |Pages: 20
ISSN: 1546-2234|EISSN: 1546-5012|EISBN13: 9781522583714|DOI: 10.4018/JOEUC.2020100103
Cite Article Cite Article

MLA

Wang, Xuan, et al. "A Predictive Analytics Approach to Building a Decision Support System for Improving Graduation Rates at a Four-Year College." JOEUC vol.32, no.4 2020: pp.43-62. http://doi.org/10.4018/JOEUC.2020100103

APA

Wang, X., Schneider, H., & Walsh, K. R. (2020). A Predictive Analytics Approach to Building a Decision Support System for Improving Graduation Rates at a Four-Year College. Journal of Organizational and End User Computing (JOEUC), 32(4), 43-62. http://doi.org/10.4018/JOEUC.2020100103

Chicago

Wang, Xuan, Helmut Schneider, and Kenneth R. Walsh. "A Predictive Analytics Approach to Building a Decision Support System for Improving Graduation Rates at a Four-Year College," Journal of Organizational and End User Computing (JOEUC) 32, no.4: 43-62. http://doi.org/10.4018/JOEUC.2020100103

Export Reference

Mendeley
Favorite Full-Issue Download

Abstract

Although graduation rates have interested stakeholders, educational researchers, and policymakers for some time, little progress has been made on the overall graduation rate at four-year state colleges. Even though selective admission based on academic indicators such as high school GPA and ACT/SAT have widely been used in the USA for years, and recent statistics show that less than 40% of students graduate from four-year state colleges in four years in the US. The authors propose using an ensemble of analytic models that considers cost as a better form of analysis that can be used as input to decision support systems to inform decision makers and help them choose intervention methods. This article uses ten years of data for 10,000 students and applies ten analytical models to find the best predictor of at-risk students. This research also uses the receiver operating characteristic curve to help determine the most cost-effective trade-off between false positive and false negative levels.