Predicting Academic Success of Students Through Educational Data Mining

Abstract

This study uses four well-known machine learning methods—Lassis, Random Forest, AdaBoost, and Support Vector Regression—that are good at making predictions. It does this to look into the accuracy of different models and groups of variables. For this study, the China Education Panel Survey statistics were used. This survey took place between 2013 and 2015. The level of effort put in by each group has different effects on academic performance. Parents' strict expectations are the most important individual predictor of academic performance. School effort has a bigger effect on academic achievement than parental and student effort when different social background factors are taken into account. There are also big differences between male and female junior high school students in China, with school effort having a bigger effect on academic achievement. The results have big effects on strategies that are meant to get students to work harder in school. They show that treatments that are tailored to each gender are needed to help all kids do better in school.