Comparative Analysis of Data Mining Methods in University Admissions

As higher education institutions increasingly adopt data-driven approaches to improve admissions outcomes, the role of Data Mining (DM) and Machine Learning (ML) in supporting equitable, efficient, and strategic decision-making has expanded significantly. This study presents a comprehensive comparative analysis of nineteen studies that applied DM techniques to university admissions. Through a structured literature review, the research categorizes the modeling approaches into six main types: classical classification and regression models, ensemble and stacked learning models, interpretable decision tree-based systems, SVM and hybrid SVM frameworks, deep learning models, and geodemographic or behavioral modeling. The analysis synthesizes model performance, feature relevance, interpretability, and fairness considerations across these categories. Findings indicate that while ensemble and deep learning models often achieve superior predictive accuracy, interpretable models such as decision trees and logistic regression remain essential in contexts demanding transparency and stakeholder trust. Furthermore, the integration of socio-demographic and behavioral data is gaining traction as a means of enhancing inclusivity, though it raises ethical concerns regarding fairness and bias. The study concludes with strategic recommendations for institutions and researchers, emphasizing the need for hybrid modeling, contextual alignment, fairness diagnostics, and validation across diverse educational settings.