Predicting Students’ Performance Using Feature Selection-Based Machine Learning Technique

Early evaluation of the students’ performance to determine their strengths and weaknesses helps them perform better in examinations. Improving students’ overall learning experiences and academic success has been a hot issue recently. In this paper, classical machine learning algorithms like the random forest, J48, and Logistic Model Tree are built and trained on student data to predict students’ performance. To improve the accuracy of the models, feature selection algorithms like correlation-based feature selection, information gain ranking filter, gain ratio feature evaluator, and symmetrical uncertainty ranking filter are used, and selected features are trained on the model and compared the performance of the models with each other.