This paper explores how to effectively predict and analyze students’ Civics course grades and their correlation with students’ behaviors based on the integrated transfer learning strategy to improve the teaching quality of Civics courses for college students through the integrated transfer learning strategy. The feature engineering method is used to extract the achievement features, and the achievement analysis and prediction model is established by Stacking integrated learning method combined with various algorithms such as KNN, linear regression and decision tree. By predicting and analyzing the Civics grades of college A students, the results show that the average error between the expected grades and the measured grades of computer science majors is 4.00 points. In contrast, biological science majors underestimate it by 5.41%. In addition, the cluster analysis of students’ behaviors using the k-mediods algorithm revealed significant differences between students with excellent academic performance and those with poorer performance in behaviors such as frequency of consumption and number of books borrowed. The integrated transfer learning strategy has a better application effect in the prediction and analysis of the performance of the Civics course, which can provide a powerful support to improve the quality of teaching.