Device Action Prediction Based on K-means and Apriori for Smart Home

This paper proposes to use K-means and Apriori to prediction device action based on time in Smart Home System. In the existing methods, the system provides services to human when conditions are met, such as high temperature and low humidity. However, those methods ignoring the correlation between device action and human activity time. In order to solve this problem, we combine K-means and Apriori algorithm to generate association rules with time information. First, calculate the appropriate K-value for K-means. We select consecutive data in weeks. The data contains the time information of device action. Second, run K-means to classify based on device action time information, and run Apriori to mine association rules based on classification. Finally, decode the associate rules and realize device action prediction. In our work, K-means experiments show that clustering according to the difference of working days is an effective way to process user time information; Apriori experiments proved that the association rules mined with the results of K-means can predict device actions in the Smart Home environment.