Enhancing Academic Performance of Rural Students in Urban Universities Using Machine Learning Classifier

Abstract

Rural students in urban universities often struggle academically because of differences in resource availability, access to learning devices, and socio-economic factors. This study focuses on detecting the academic progress of these groups through the application of machine learning classifiers as early intervention for improving their scores. In this research work, we plan a holistic methodology which incorporates logistic regression for preprocessing the data, Chi-square based feature selection for dimensionality reduction and Remora Optimized MK-SVM for classification process. Chi-square test helps to identify the highly informative features in a pool of the given variables, to build a predictive model with lesser complexities to improve accuracy and reducing overfitting. By introducing yet another level of optimization through MK-SVM with Remora, which is a nature-inspired optimization technique, researchers can enhance the classifier’s potential to capture underlying complex patterns embedded in the student data. In addition, this study adopts the train-test split mechanism, so that the evaluation of the model can be more rigorous in examining the generalization capability of the proposed system. The outcome of our study reveals that the integration of logistic regression for preprocessing and feature selection with the Remora-optimized MK-SVM achieves a substantial accuracy boost over classical approaches, suggesting actionable guidance for potential strategies to enhance the academic success of rural students in urban higher education institutions along with higher accuracy percentage of 99.5. This study offers a new methodology in the field of educational analytics, illuminating how machine learning tools can help close educational gaps.