Analyzing Supervised Learning Models for Predicting Student Dropout and Success in Higher Education

The global education industry has experienced significant transformations, yet it continues to grapple with challenges, particularly in higher education, such as declining student success rates and course abandonment. Addressing these issues necessitates proactive identification of students at risk of failure and timely intervention, through appropriate models. This study focuses on a comparative analysis of various supervised learning models that effectively predict student success and dropout. Specifically, the performance of five models, namely MLP (Multilayer Perceptron), SL (Simple Logistic), DT (Decision Tree), RF (Random Forest), and REPTree (Reduced Error Pruning Tree), is evaluated using a Kaggle dataset comprising 35 attributes and 4424 instances. The experiment encompasses all attributes and evaluates model accuracy based on Precision, Recall, and F-measure for all 5 models. Additionally, the study also compares Correctly and Incorrectly Classified Instances of these Machine Learning models. The findings reveal that Random Forest achieves the highest percentage of correctly classified instances and surpasses other supervised learning methods in terms of accuracy.