Optimizing Student Performance Prediction: A Comparative Analysis Using Machine Learning

The analysis of student performance is a data-driven process. This analysis helps to provide high-quality education, a strategic way to select quality students, predict a student's future, etc. A highly competitive and complex environment is observed due to the increase in the number of institutions and the large number of specifications in the educational area. In that scenario, the analysis of student performance faces the challenge of achieving high accuracy in examining factors like demographics, behavior, and academics for a student. We have observed that the regression technique in machine learning helps us solve this challenge. In the proposed work, we have analyzed the student performance using various regression techniques such as linear regression, lasso regression, and SVM regression. In the comparative analysis, we observed that linear regression is highly effective in real-time applications, whether the lasso regression can manage the overfitting through regularization or SVM regression can take care of high-dimensional data. In the proposed work, the maximum accuracy (98.20%) is achieved in the ANN technique, which is higher than other existing techniques. The comparative study is also shown in the results section of the paper.