Prediction of Student’s Academic Performance Using Learning Analytics

Academic performance is the assessment of knowledge gained by students for a particular period. Therefore, predicting student’s academic performance is a vital part of quality assurance in higher learning. As such, their prediction becomes pertinent to higher institutions as it can be used to monitor student’s progress and forestall the risk of students derailing from their academic paths. This study proposes a hybrid of linear regression and a k-means clustering model to predict student’s academic performance. (The gain of these models is that useful and interesting patterns can be discovered from the data provided)? The proposed hybrid model has been tested on the Covenant University, Nigeria, student dataset. The main features of the dataset include department, gender, Secondary School Grade Point Average (SGPA), Cumulative Grade Point Average (CGPA) for each academic session, and final CGPA. The linear regression model results showed improved performance with a coefficient of determination (R2), Mean Squared Error (MSE), and F-statistic value of 0.9842, 0.00077, and 10448.24, respectively. In contrast, the Clustering model showed good measures with between-cluster error, within-group error, and variance of 462.27, 197.12 and 0.70, respectively. This new hybrid approach displays expressively improved performance. The analysis of the prediction performance indicates that the proposed ensemble scheme performs well. At the same time, the percentage of correctly classified instances is increased as new performance attributes are added during the academic year. Additionally, the investigation was carried out on how soon predictions could be made to offer a timely intervention and enhance students’ performance.