Abstract: Based on the analysis of electric pressure cooker user behavior data, it is clear that user habits have a strong regularity. Further clustering analysis of users reveals that there are large differences between different types of users, and the gap is further increased when it comes to individual users. Through the time nodes of the local electric pressure cooker usage data, and the information of function options and setting parameters, use the machine learning model XGBoost and statistical analysis knowledge to carry out multi-dimensional data mining, and interpret the user usage habits of electric pressure cookers, and combine with the existing iot function, and complete the data of individual users through the user's historical reservation time, use function content, etc. The data of users' historical reservation time and function content can be used to remind individual users of the reservation and the intelligent recommendation and prediction of users' function parameters, so as to strengthen the intelligent function of electric pressure cooker for the guidance of users' use behavior and further enhance the intelligent interaction between kitchen appliances and users.

Key words: Big data analytics, Machine learning, XGBoost, Intelligent prediction, Intelligent interaction