In the context of education big data, it uses data mining and learning analysis technology to accurately predict and effectively intervene in learning. It is helpful to realize individualized teaching and individualized teaching. This research analyzes student life behavior data and learning behavior data. A model of student behavior characteristics is constructed, and a robust multi-task learning method is used to construct an academic prediction model. According to the prediction results, different intervention measures are taken for students with academic excellence and academic difficulties. Finally, it takes the one-semester blended teaching course of a certain university as an example. The research results show that in terms of predictive models, through the analysis of student behavior characteristics data, the model can accurately identify the learning status of students. In terms of intervention, it can play a positive role in promoting students with high learning and can effectively promote students with learning difficulties.