PREDICTING STUDENTS PERFORMANCE ACADEMIC AND CORRELATION ANALYSIS NEURAL NETWORK ALGORITHM

ABSTRACT The accurate prediction of student academic performance is a critical component in enhancing educational planning, resource allocation, and individualized learning support. This study aims to predict student academic performance and perform correlation analysis using a neural network algorithm. Guided by the Object-Oriented Analysis and Design Methodology (OOADM), the research pursued four objectives: collecting and preprocessing multi-source student datasets, developing a neural network model to predict performance, analyzing the correlation between academic outcomes and JAMB scores, and benchmarking the neural network against traditional machine learning algorithms. The implemented neural network model demonstrated a predictive accuracy of 98%, substantially outperforming the decision tree model, which recorded 74.2%, and the logistic regression model, which achieved 68.5%. This represents a performance improvement of 15.2% and 20.9%, respectively, underscoring the neural network’s superior ability to capture nonlinear relationships and complex academic patterns. Error analysis further validated these findings, with the proposed system attaining a Mean Squared Error (MSE) of 0.041 compared to 0.087 for the decision tree and 0.104 for logistic regression. Beyond predictive modeling, the correlation analysis module revealed a strong positive relationship between class attendance and GPA, with a Pearson coefficient of +0.76. This result aligns with existing systems, who reported a coefficient of +0.71 using traditional statistical methods. The integration of automated visualization and reporting tools enhances interpretability, enabling educators and administrators to make data-driven decisions with greater confidence. Overall, the study demonstrates that neural networks not only provide more accurate and reliable performance prediction than traditional models but also offer valuable insights into academic determinants such as JAMB scores and attendance. The findings contribute to the growing body of knowledge on AI-driven educational analytics and highlight the potential of intelligent systems in advancing student success strategies.