Do Machine Learning Algorithms, with their Potential to Significantly Improve the Accuracy of Predicting Student Success, Offer a Promising Future for Education?

Academic institutions considers factors including student success, graduation, degree completion, and improved retention as of vital importance. Educational institutions understand the complexity of accurately predicting factors of student persistence. In this study, we identify and assess multiple student success factors (database of 19,000+ students) at an academic institution. For analysis, we deploy different data mining classifiers (e.g., Logistic Regression (LR), Support Vector Machine (Linear Classifier), Support Vector Machine (Radial Basis Function), Decision Tree (DT), Random Forest (RF), K-mean Nearest Neighbors (KNN) to examine student success scenario. Further, we report descriptive statistics of student demographics (e.g., age, registered credit, gender, race, GPA, etc.) along with applying the Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance in ML datasets and improve model performance. Key results indicated student success factors are credits completed, first semester grade point average (GPA), age, and registered credits. As per our study findings, RF algorithms works best and achieved 86% accuracy. We also report six evaluation metrics known as Accuracy, Sensitivity, Specificity, Precision, F1-score, and Kappa to assess model performance. The practical implications of this research is that similar academic constituencies can launch preventive actions to minimize dropouts and gain student persistence, thereby enlightening the academic community about the potential of machine learning in this context.