Construction of Teaching Effect Evaluation Model of Ideological and Political Education and the Marxism in China Based on Big Data

This study develops a big data–driven model to address limitations in evaluating the teaching effectiveness of ideological and political education and Marxism in China. The model integrates multisource data from theoretical cognition, value identification, practice transformation, and learning behavior characteristics. Through quasi-experimental validation, the model demonstrates strong reliability (composite reliability > 0.80, average variance extracted > 0.50). Dynamic intervention significantly improves teaching effectiveness (local average treatment effect = 0.29–0.49). By combining eXtreme Gradient Boosting and Shapley additive explanations, the model enhances the interpretability of artificial intelligence technologies, accurately identifies different subgroups of students with learning difficulties, and shortens response time. The proposed evaluation model significantly outperforms random forest and logistic regression models across all indicators. The model establishes a data-driven paradigm for improving the quality of ideological and political education.