A Hybrid Weight based Feature Selection Algorithm for Predicting Students’ Academic Advancement by Employing Data Science Approaches

Abstract

PerformanceX is a proposed system that combines Educational Data Mining (EDM) techniques to enhance student performance and reduce dropout rates. It employs a hybrid feature selection approach to identify the most significant attributes from student academic datasets, eliminating unnecessary features that are not crucial for predicting performance. The selectX algorithm, a critical component of PerformanceX, selects a limited number of high-performing features to optimize student learning effectiveness and prediction accuracy. The system applies various machine learning classifiers, including a fusion Voting Classifier, to different subsets of features, ultimately determining the best combination. The study achieved an impressive accuracy rate of 99.41%, with the selectX approach utilizing 10 features in conjunction with a random forest (RF) classifier offering the highest accuracy. These findings underscore the importance of categorizing student performance based on a concise yet meaningful set of features, leading to improved student quality and career progression. The research value of PerformanceX lies in the development of a performance forecasting system that eliminates irrelevant information and provides precise predictions for student performance. Its efficacy and efficiency make it an invaluable tool for educators and educational institutions. By assisting students in selecting appropriate courses to enhance their performance and advance their careers, PerformanceX contributes to diminishing dropout rates while fostering positive student outcomes.