Student Performance Prediction in Learners Centric Approach with Machine Learning

Predicting student performance helps educators locate students who are at risk and tailor appropriate and timely interventions for those students. This research proposes a learner-centered machine learning framework to integrate demographic, academic and behavioral features in order to predict student grade performance. The dataset consists of 2392 students and 15 attributes including age, gender, parental education, study time, absences, and extracurricular activities. Four supervised learning models - Logistic Regression, Decision Tree, Random Forest and Support Vector Machines (SVM) were trained and measured using 70:30 stratified split. The performance of the model was evaluated using accuracy, precision, recall, and F1-score metrics. Among these, Decision Tree classifier achieved the highest accuracy (92.48%) which was followed by Random Forest (88.31%), SVM (83.51%) and Logistic Regression (75.16%). The results show that such factors as study time, absences, and parental involvement were the most predictive. The proposed learner-centered approach shows that the combination of contextual, behavioral, and academic data can greatly increase the predictive accuracy and the interpretability of the data, facilitating early risk detection and intervention in education.