ARTICLE IN PRESS JID: EOR [m5G;April 13, 2019;11:38] European Journal of Operational Research xxx (xxxx) xxx Contents lists available at ScienceDirect European Journal of Operational Research journal homepage: www.elsevier.com/locate/ejor Development of a Bayesian Belief Network-based DSS for predicting and understanding freshmen student attrition Dursun Delen a,∗ , Kazim Topuz b , Enes Eryarsoy c a Department of Management Science and Information Systems, Spears School of Business, Oklahoma State University, Stillwater, Oklahoma, USA b School of Finance, Operations Management, and International Business, Collins College of Business, The University of Tulsa, Tulsa, Oklahoma, USA c Department of Management Information Systems, School of Management and Administrative Sciences, Istanbul Sehir University, Turkey a r t i c l e i n f o Article history: Received 21 February 2018 Accepted 7 March 2019 Available online xxx Keywords: Student retention Prediction Elastic net Bayesian Belief Network (BBN) Imbalance data a b s t r a c t Student attrition – the departure from an institution of higher learning prior to the achievement of a degree or earning due educational credentials – is an administratively important, scientifically interest- ing and yet practically challenging problem for decision makers and researchers. This study aims to find the prominent variables and their conditional dependencies/interrelations that affect student attrition in college settings. Specifically, using a large and feature-rich dataset, proposed methodology success- fully captures the probabilistic interactions between attrition (the dependent variable) and related fac- tors (the independent variables) to reveal the underlying, potentially complex/non-linear relationships. The proposed methodology successfully predicts the individual students’ attrition risk through a Bayesian Belief Network-driven probabilistic model. The findings suggest that the proposed probabilistic graphi- cal/network method is capable of predicting student attrition with 84% in AUC – Area Under the Receiver Operating Characteristics Curve. Using a 2-by-2 investigational design framework, this body of research also compares the impact and contribution of data balancing and feature selection to the resultant pre- diction models. The results show that (1) the imbalanced dataset produces similar predictive results in detecting the at-risk students, and (2) the feature selection, which is the process of identifying and elim- inating unnecessary/unimportant predictors, results in simpler, more understandable, interpretable, and actionable results without compromising on the accuracy of the prediction task. © 2019 Elsevier B.V. All rights reserved. 1. Introduction Predicting attrition (i.e., early identification of undesirable de- partures) has always been an intriguing and challenging problem for researchers and decision makers (i.e., practitioners, administra- tors or business managers). Attrition analysis—as a tool to better understand and manage retention—is an important subject in a variety of domains and based on the specific domain, it may be named differently. For instance, in marketing, it is often called "customer churn analysis" referring to timely identification of the at-risk customers (i.e., the ones about to leave you—your products and/or services—for your competitors’). The common marketing philosophy that states “acquiring a new customer is ten times as costly as retaining current customers” is a good testament to the importance of predicting and properly managing customer churn. Attrition is also an important concept in human resource ∗ Corresponding author. E-mail addresses: dursun.delen@okstate.edu (D. Delen), kat0141@utulsa.edu (K. Topuz), eneseryarsoy@sehir.edu.tr (E. Eryarsoy). management, specifically in employee retention, where accurate prediction and management of at-risk employees would save time and money to the organization in maintaining a productive and capable workforce. In this study, the focus is on student attrition, where better understanding and accurate prediction of attrition can lead to effective management of retention and graduation of students in higher education. Student attrition is the precursory dropout of students in institutions of higher learning prior to achieving any recognized degrees or credentials (Johnson, 2012). Decreasing the attrition rate (i.e., increasing the retentions rate) provides an institu- tion a better chance for procuring a higher status in college evaluations/rankings, potentially leading to increasing funding opportunities, recruiting better students, and having a less com- plicated path to program/degree accreditations. In addition to these explanations, financial loss, increased federal and state level attention make administrators in institutions of higher learning feel continually increasing levels of pressure to create and execute strategic initiatives to decrease student attrition (Golde, 2005). Nowadays, a vast majority of higher education institutions have "student success centers" and related programs and services https://doi.org/10.1016/j.ejor.2019.03.037 0377-2217/© 2019 Elsevier B.V. All rights reserved. Please cite this article as: D. Delen, K. Topuz and E. Eryarsoy, Development of a Bayesian Belief Network-based DSS for predicting and understanding freshmen student attrition, European Journal of Operational Research, https://doi.org/10.1016/j.ejor.2019.03.037