The Statisitical Analysis of Undergraduate Grades Based on the Adaptive-Lasso

In this paper, the multivariate linear statistical method is applied to research the undergraduate grades of students from the school of mathematics in Hefei University of Technology, and explore the impact on the later achievement by the early stage of achievement from all undergraduate courses. First, we get the main components from the previous courses by principal component analysis, then construct a linear regression model between the later achievement and main components by the stepwise regression method. Next, a linear regression model between the later achievement and the early stage of achievement from all undergraduate courses is constructed by Adaptive-Lasso method. Finally, comparative analysis is performed for the result of the above models. The research shows that the principal component regression model based on the Adaptive-Lasso method can well fit the later achievement, and give a reasonable explanation for the later academic performance.