Abstract
This paper adopts the factor model to forecast the power demand of the residential and service sectors, comparing its forecasting performance with the VAR model. The factor model uses the principle component analysis, generating factors from abundant information to represent the complicated data. We find that the factor model such as IMS, UN, and DMS outperforms the forecasting performance of the vector autoregressive model. As a result, the factor model has more advantage in forecasting the power demand of service and residential sectors than the VAR model.