A large amount of learning behavior data is generated in online course education, and the correlation between these data can be effectively mined by using data mining and behavior analysis techniques. In collaboration with higher vocational schools and enterprises, this paper develops and enhances the modular business English boutique online open course on the SPOC teaching platform. Relying on the business English boutique online open course, we are exploring an effective new teaching mode. Clustering analysis of online learning behavior using K-Means algorithm, dividing learning status, achieving short-term monitoring and evaluation of online learning behavior. Investigate the potential relationship between grades by mining the association rules of grades in two semesters using the Apriori algorithm. To study the changes in students’ learning effects and motivation under the new teaching mode, qualitative and quantitative analyses are utilized. The experimental analysis shows that there are 2 peaks in the distribution of students’ learning hours during the learning process of the SPOC “Business English” course: fragmented learning within 0~5min and classroom-based learning around 41~45min. By using data analysis, it is possible to gain a more accurate understanding of their learning situation and develop effective learning programs.