PREDICTING STUDENTS' ACADEMIC PERFORMANCE USING MACHINE LEARNING ALGORITHMS AND CORRELATION ANALYSIS

The aim of this research was to develop a predictive system capable of analyzing and forecasting student academic performance using machine learning techniques, with the ultimate goal of enhancing educational outcomes. While traditional statistical models provide limited insights, this study sought to combine correlation analysis with the Random Forest algorithm to create accurate and interpretability in academic performance prediction. The method involved collecting and preprocessing student data from past academic sessions. Correlation analysis was used to identify the most influential predictors of performance. The resulting dataset, which contained attributes highly correlated with final academic outcomes, was then used to train a Random Forest classifier. A Random Forest classifier was then trained and tested using the prepared dataset, demonstrating high accuracy and robustness. A Streamlit dashboard was also developed to enable educators to visualize correlations, analyze predictions, and implement interventions for at-risk students. The results revealed that the Random Forest model achieved an accuracy of 90\%, outperforming baseline statistical models in predicting student academic outcomes. Key predictors such as attendance and prior academic performance significantly influenced final outcomes. The system provided actionable insights that can be used to monitor, identify, and intervene with students who are struggling academically. Based on these findings, the study recommends that academic institutions adopt predictive analytics systems for early warning, personalized monitoring, and intervention. It further suggests extending the model to incorporate psychological and socioeconomic factors to strengthen the predictive power and practical impact of such systems. ​Keywords: Student Academic Performance, Machine Learning, Correlation Analysis, Random Forest, Predictive Analytics, Educational Outcomes