Tech Blog: Cat Tracks; How One UA Professor is Changing Student Retention with Big Data

Each year in the United States most public universities retain an average of 75 percent of new students, a challenge that has been ongoing for over forty years now. At the University of Arizona, where around 10,000 students join the university each fall, this equates to about 2,500 students dropping out. While traditional methods of prediction have suggested using students’ grades as an indication of their likelihood to drop out, this has proven unsuccessful over time.

UA faculty member Sudha Ram, Ph.D. is tackling this challenge by repurposing big data to identify students’ social interactions and predict their likelihood of dropping out. Ram currently serves as the Anheuser-Busch Chair in MIS, Entrepreneurship and Innovation, a professor of MIS, and Director of the INSITE Center for Business Intelligence and Analytics(link is external), where she implemented machine learning algorithms to analyze students’ university smart card transactions; the goal was to analyze the social networks and sequences of locations to infer a student’s integration into the university environment.

“If you think about it, their smartcard is like a series of check-ins at different locations and it has a timestamp. So we thought, ‘what if we could anonymize this data and for each student just get a series of check-ins with a timestamp and location? Maybe then we can construct networks of interactions among them.’ And then we also used the same data to see if they’ve established a regular pattern or daily routine on campus.”

Ram’s research, which included analyzing about 12 weeks of anonymous UA student CatCard data, resulting in an 85 percent rate of accuracy in predicting dropouts. By implementing this technology into their student retention practices, universities can identify at-risk students and intervene within the first semester before their decision to dropout has been made.

While UA is currently experiencing its highest freshman retention rates yet (greater than 83 percent), Ram’s vision is that other sectors outside of higher education will reap similar benefits from this technology, saying “there are a variety of fields in which we feel our techniques could be extended and adapted for prediction purposes” [1]. These include healthcare, customer and employee retention/satisfaction, to name a few.

Learn more about this technology:

UA16-138 Predicting Student Retention Using Smartcard Transactions

Learn more about other big data-related technologies from the University of Arizona:

UA13-024 Health Analytics and Cyber Security Software – Asymmetric Threat Analysis Tool (“Atrap”)

UA17-168 Online Visualization Engine for University International Data

References:

[1] University Communications. “UA Retention at an All-Time High.” UANews, The University of Arizona, 19 Feb. 2018, uanews.arizona.edu/story/ua-retention-rates-exceed-abor-goals.