Mining Educational Data to Create a Model to Predict Student Retention

Abstract

Student retention is a widespread issue in higher education. This study applies data mining methods to analyze student data at MTSU, specifically focusing on minority student groups. The hope of this study is to determine a set of attributes that are highly predictive of student retention and to develop models to predict the retention status of future students within these target groups. Decision tree classification models will be created for each target minority group in order to predict the retention status. We found that a student’s GPA and financial factors are the most predictive on the student’s retention status. The developed models were able to successfully classify students at rates as high as 68.7%. We hope that this data can help to provide the university with a way to identify students with a high risk of failing to remain enrolled and to improve retention rates of minority students.