A Data-Driven Approach to Aid in Student Retention at Higher Education Institutions

• In Event: AERA Virtual Poster Session
  In Virtual Poster Session: SIG-Data-Driven Decision Making in Education Virtual Poster Session

Abstract

Student retention is a major challenge at American universities with the average six-year graduation rate hovering around 59%. This paper takes a novel approach to the issue of identifying at-risk students by using “Did Not Register” (DNR) lists containing data on students yet to registered for the upcoming semester. Using three years (2019 – 2021) of data we demonstrate that standard machine learning models like XGBoost can achieve a high degree of accuracy (96%) in predicting students likely to be in the DNR list. Institutions can deploy these models as a data driven decision making tool to enrollment management, student advising, administration, and faculty to help in not only increasing retention rates, but also better planning and budgeting.

Authors

• J.D Jayaraman, New Jersey City University
  Presenting Author

• Ruth Ortiz, New Jersey City University
  Non-Presenting Author