Prediction of Student Academic Performance Based on Machine Learning Model

The performances of students are one of the major factors that contribute to the growth, relevance, and marketability of any tertiary institution. Students’ performances can be predicted and enhanced through their attitudes to lectures, and other factors such as health status, financial ability, and study environment. Students’ attitudes towards classes, tests, examinations, and overall performances can be enhanced by analyzing and predicting previous records. In a higher institution of learning, analysis of students’ performances is important to improve class attendance, reduce failure rates, and determine the output of students and lecturers after each semester or session. Therefore, this project aims to enhance students’ performance in higher institutions by assessing previous performances and factors contributing to learning outcomes. Factors such as lecture durations, lecture time, attendance, gender, and mode of learning were studied and analyzed to predict students’ performances. To achieve this, linear regression was utilized to analyze and predict student performance based on data extracted from existing student records. By training and testing the model with this data, the study aims to provide valuable information and recommendations to students and lecturers to enhance educational outcomes.The findings underscore the importance of analyzing and leveraging factors contributing to student success, offering valuable insights for both students and lecturers to improve educational outcomes. The performance of the proposed study was assessed using evaluation metrics. The Root Mean Squared Error (RMSE) yielded an average error of 1.6429, indicating the model's predictions have a relatively small deviation from the actual values. The Mean Absolute Error (MAE) resulted in an average error of 1.14043, further highlighting the model's accuracy in predicting the target variable. Additionally, the R-squared accuracy of 88% demonstrates the model's ability to explain a significant proportion of the variance in the data. Overall, the accuracy of the model is 88.3%.