A comparison of different machine learning methods in the prediction of college readiness

This study employs a machine learning approach for students’ college admissions. Specifically, it utilizes five machine learning models, including Linear Discriminant Analysis (LDA), K-Nearest Neighbor (KNN), Decision Tree, Random Forest, and Naïve Bayes Model, to predict students' admissions. Using the ACT dataset, this study presents analyses of machine learning models employing five different models. The findings indicate that all five models perform well overall, with the Random Forest model demonstrating the highest accuracy. The bootstrapping results illustrate that all five models are robust, exhibiting small standard deviations. While the accuracy rankings change slightly with different cut scores, the best- and worst-performing models remain consistent across cut scores. In general, the ACT score is the most significant factor in college admission. However, discrepancies exist between the model-generated predictions and the weights determined by the operational RAI index. This study seeks to enhance the admissions process by applying machine learning models, offering a more accurate and robust approach through predictive modeling.