Research on the Application of an Improved Machine Learning Algorithm Based on Weighted Grey Projection in Educational Data Mining

This paper introduces the application of machine learning models in educational data mining, examining the performance of support vector machines, logistic regression, random forests, and multilayer perceptrons in predicting student learning outcomes. It also introduces the weighted grey relational projection method to improve the prediction models. By filtering similar sample sets for training machine learning models based on the degree of grey relationality among samples, the paper analyzes the optimization effects of this improved algorithm on different models. The results indicate that the predictive performance of the improved models has indeed been enhanced, with the weighted grey relational projection method showing the most significant optimization effect on support vector machines.