This paper describes in detail the commonly used educational data mining algorithms and proposes a new data mining-based teaching model for aesthetic education in modern universities. Data mining technology is used to collect students’ online learning behavior and learning performance data, and collate them to lay the foundation for improving personalized teaching work and teaching resource management in aesthetic education. To evaluate the effectiveness of the application of the modern education model implemented in colleges and universities, the rough set evaluation method was also introduced. The evaluation scores for the modern education model of the aesthetic education program in higher education are all above 0.8, and the rating scale is exceptional, as shown in the results. Compared with the previous stage, the students’ learning behavior input and knowledge mastery in the latter stage have been significantly improved, and the average score of declarative knowledge has increased by 27.42. The curriculum resources for calligraphy, drama, film, and television should be improved.