Three data mining models suitable to data processing have been proposed according to some technical challenges on the characteristics data for Internet of Things (IoT). The first model is the multilayer data mining model, which consists of data collection, data processing, event processing and data mining service. The second model is distributed data mining model that could be used to solve the problem of data stored in different locations. The third model is grid-based data mining model, which could utilize the potentially unlimited amount of data by using grid. At the same time, the paper learns from \"Decomposition - build \"theory to solve the data mining problem in the internet of things. Finally, the key issues are discussed about model application.