Elevating Student Success: Harnessing Machine Learning to Enhance University Completion Rates

Abstract: This research paper presents a machine learning approach designed to aid universities in identifying students at risk of not completing their studies. Predicting student attrition and academic success is pivotal for universities to proactively intervene and enhance student retention rates. The proposed machine learning model harnesses historical student data, encompassing demographic information, academic performance, and financial status, to construct predictive models. These models employ a comprehensive array of algorithms, including Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), Logistic Regression (LR), Decision Tree (DT), and Feedforward Neural Network (FNN), to categorize students into distinct retention-completion groups. By adopting this approach, universities can effectively allocate resources and implement targeted interventions, offering support to students likely to either transfer out or face academic challenges. In pursuit of these objectives, this paper highlights the specific methods employed to gather and preprocess historical student data. The rationale behind the selection of each algorithm is elaborated, showcasing their combined efficacy in providing a holistic analysis of student retention patterns. As an embodiment of data-driven education, this research holds the potential to reshape how universities approach student retention. Beyond the immediate insights derived, this work suggests a promising trajectory for further research and seeks to uplift academic outcomes and foster a more supportive learning environment.

Keywords: Machine Learning, Decision Trees, Feedforward Neural Network, Deep learning, Retention Rate, Student Success

DOI: 10.54941/ahfe1005822

Cite this paper:

​