Using Seasonal Time Series Analysis to Predict China´s Demand of Electricity

Time series analysis is to explain correlation and the main features of the data in chronological order by using appropriate statistical models. Since the past electricity generated sequence in China shows a strong seasonal variations and several values for January are lost in recent years, estimating the missing values is an important task before building a model. This paper will estimate the missing values by Holt-Winters exponential smoothing containing seasonal components, and then select appropriate multiplicative seasonal ARIMA (Integrated Autoregressive Moving Average) model to simulate the complementary electricity generated series and forecast the demand in the next two years. Building models and predicting the future values for the electricity series can not only help estimate the development of China´s electricity market but also provide credible scientific basis for policy makers to formulate the electricity production indicators.