1. User behavior collection. What kind of data is useful, even scholars with years of data mining experience are difficult to assert. The current common practice in the Internet industry is to store as much data as possible. For example, Taobao, all users' login time, login location, browsing products, mouse behavior, web page stay time, collection of products and so on are recorded completely. With enough comprehensive data, you can analyze the data from multiple perspectives and explore the value of the data. Catering industry also needs to record the user's historical behavior on their own platform, such as which dishes they have browsed, where they have stayed for a long time, and what dishes they have ordered at what time, in what place and by whom. This data will lay the foundation for analyzing user behavior, understanding user preferences, and so on.

2. Merchant operation data collection. The data from all the merchants will be centralized, and from the perspective of data warehousing, the frequent item sets will be mined, sales trends will be analyzed, and the temporal, spatial, and group differences in sales will be understood, which is of great significance for the merchants to understand the laws of the industry, improve the business model, and enhance the sales of the merchants.