Home Energy Management System, HEMS for short, is expected to conserve electric power. In order to realize appropriate energy management and device control by HEMS, it is important to predict future behavior of electricity consumption. Typically, the method of future prediction of electricity consumption is built by using a combination of machine learning and environmental data which is acquired by sensor network. Considering to apply the method to HEMS, we point out that it is difficult to build such a network into a general residence. We thus propose a method to use network-connected home appliances as a sensor network. Environmental data is collected by the network which is composed by ECHONET Lite protocol. Finally, our system outputs models for future prediction of electricity consumption by using machine learning.