关键词: CPI, Holt-Winters模型, 季节因素, 预测

Abstract: CPI are relation to people’s daily lives, but also an important indicator of economic analysis and decision-making, price monitoring and regulation, national economic accounting. In the paper, we take China’s clothing, food, housing, transportation and urban & rural fixed time classified monthly CPI data from Jan 1997 to Dec 2009 for samples. On the basis of analyzing their fluctuation characteristics, by exponential smoothing method of Holt-Winters model they will be decomposed into seasonal and trend volatility. The results show that classified CPI have a clear trend and seasonal characteristics differently, and further get the arrival time of peaks and troughs. Model has very good fitting results for them. The MAPE are 0.253%、0.816%、0.364%、0.359%、0.391%、0.338%. On this basis, we conduct a scientific prediction for classified CPI each month of China in 2010.

Key words: CPI, Holt-Winter, model, Seasonal, factors, Forecast