摘要
以我国苹果批发市场价格为研究对象,利用2006年7月7日至2012年3月30日期间的300个周数据作为分析样本,通过对时间序列的平稳性、趋势性、季节性、异方差等数据特征进行统计检验,筛选出双指数平滑模型、Holt-Winters乘法模型、ARIMA(1,1,4)模型为我国苹果市场价格短期预洲的适用模型,以此为基础,以误差平方和最小为最优准则建立了组合预测模型.经对未来3期的苹果市场价格开展预测,结果表明,组合预测的精度要高于单项时间序列模型,组合预测方法完全适用了农产品市场价格的短期预测.
Taking the apple price of wholesale market in China as study object,using the data of 300 weeks from 2006-7-7 to 2012-3-30 as analysis sample,by statistic test of the data characteristics such as Smooth,Trend,Seasonality,Heteroskedasticity etc.of time series,we choose the Double exponential smoothing model,the multiplicative model of Holt-Winters,ARIMA（1,1,4） model as applicable model of short term forecast of apple market price in China.Based on these results,we establish the Combination Forecasting Model with the least sum of error square as optimal criteria.By carrying out the forecast of apple market price in future three periods, the results show that the accuracy of combination forecasting is higher than individual time series model,and the combined forecasting method is fully applicable to the short-term forecasts of agricultural products market price.
出处
《系统科学与数学》
CSCD
北大核心
2013年第1期89-96,共8页
Journal of Systems Science and Mathematical Sciences
基金
国家"十一五"科技支撑项目"果蔬市场价格短期预测系统研究与示范"(2009BADA9B05)
中央级公益性科研院所基本科研业务费专项"中国水果市场价格波动及风险度量研究"(项目编号2012-J-10)
关键词
农产品
市场价格
短期预测
组合预测
Agriculture products
market price
short-term forecast
combination forecasting
