月度数据季节因素调整和预测  

Seasonal Adjustment and Forecast for Monthly Data

作　　者：桂文林[1] 

机构地区：[1]惠州学院数学系

出　　处：《统计研究》2011年第6期79-85,共7页Statistical Research

基　　金：广东省哲学社会科学基金(编号:09E-04);广东省自然科学基金(编号:9151051501000066)

摘　　要：由于受气候条件、节假日、人们的风俗习惯、人口和国民经济增长等因素的影响,客运量呈现出周期性的增长趋势变化。为帮助客运部门更好地安排客运计划,本文通过指数平滑法中的Holt-Winters模型将时间序列数据分解为季节波动和趋势波动。并对我国铁路、民航、水运和公路的2002-2009年的客运量数据进行拟合。结果表明,铁路和民航客运量数据具有明显的线性趋势和季节性特征,并进一步得出其波峰和波谷到达的时间;模型对铁路、民航、水运和公路客运量均有非常好的拟合效果,其平均绝对百分比误差(MAPE)依次为5.536%,7.49%、6.070%和3.633%。在此基础上对我国2010年各月份的客运量进行了科学预测。Due to weather conditions,holidays,people＇s customs,population and economic growth and other factors,the volume of passenger traffic shows a cyclical growth trend.To make better plan for the passenger transport sector,in this paper,we use Holt-Winters model of the exponential smoothing method to decompose the time series data for seasonal and trends fluctuation,and use the monthly data of the volume of China＇s railway,road,river and airplane passenger traffic during 2002 to 2009 to fit.The results show that railways,airplane passenger traffic data have a clear linear trend and seasonal characteristics,and further we get the arrival time of peaks and troughs.Model has very good fitting results for them.The average percentage error（MAPE）are 5.536%,7.49%、6.070%和3.633%.On this basis,we conduct a scientific prediction for the volume of passenger traffic of each month in 2010.

关 键 词：Holt-Winters模型 客运量 预测 季节因素 

分 类 号：C812[社会学—统计学]