As teaching models increasingly integrate with the big data era, data-driven and technology-enabled approaches have become important tools for evaluating the effectiveness of ideological and political education. This paper explores reliable methods for assessing the effectiveness of ideological and political education primarily through learning analytics and behavior tracking. The research data was preprocessed by merging data, handling missing and abnormal cases, and converting features. The Gaussian mixture model clustering algorithm is selected, and a probabilistic model is introduced to handle complex distributions of various student behavior data, achieving clustering of student behavior data. Based on student learning activity sequences and cognitive styles during the ideological and political education process, a learning behavior model is constructed to analyze the quality of the ideological and political education process based on information from the learning process. Taking second-year students from a certain major at University E as the research subjects, after applying Gaussian mixture clustering to their ideological and political-related learning data, the learning behavior model proposed in this paper achieved a prediction accuracy of 0.8839 for student learning behavior and performance.