This research aims to realize a data-driven education system that recommends optimal individualized information by understanding the learner’s learning situation in each learning topic unit. We will conduct research on (1) a dense sensing method to record learning activities, (2) a deep activity analysis method to understand the context of learning activities, and (3) a persuasive feedback method to promote behavioral change. We will conduct demonstration experiments of the data-driven PDCA cycle in practical university education settings to verify the usefulness and effectiveness of our research results, and disseminate our findings widely both domestically and internationally.