Last modified: 2019-12-19
Abstract
Retention of students is one of the significant challenges the universities face. To a large extent, it depends on the ability the students to successfully pass the courses. We had anecdotal evidence that that a combination of some courses increases the chance of failure, while taking the courses in different semesters is successful. To evaluate this evidence, we applied two data analytics methods - FP-Growth and Collaborative Filtering – over an anonymized dataset which provided student aliases with academic difficulties, the semester of academic difficulty, the GPA of the student for that semester of academic difficulty, as well as the courses the students were taking that same semester, along with the grades of that student in all courses of that semester. The dataset was for the past several years.
As a result of the applied data analytical methods, coupled with a qualitative review of the results by anonymous students, we identified courses that appeared to be problematic with regard to concurrent enrollment; we additionally determined that student interest in course material played a role in doing well in some courses. The results led to a better advising plan. The applied approach could be extended to other programs and disciplines.