A Cluster Model Applied to Power Consumption

The main impact of this paper is to compare two methods of grouping; Kmean and Usage function, for promote the effect of two classifications. According to the simulation results, respectively, using those two cluster methods to divide power data into 10 groups, the result is Usage function can significantly improve the C5.0 classification and prediction, but also can enhance the classification and prediction of CART. We hope through the effective grouping method to efficient distribution and use of electricity.