基于季节指数修正因子的交乘趋向预测方法  

A Forecasting Method Based on Season Index Modify Factor

作　　者：王跃军[1] 刘万春[1] 王正平[1] 朱玉文[1] 

机构地区：[1]北京理工大学计算机科学与工程系,北京100081

出　　处：《军民两用技术与产品》2006年第4期43-44,共2页Dual Use Technologies & Products

摘　　要：季节预测模型中的交乘趋向模型虽然能根据历史趋势的变化加大季节跳动幅度,但对于季节指数的处理只是使用了历史季节指数球平均,无法很好地提取历史数据中含有的季节特征信息。在季节交乘趋向模型中,先对季节指数做出趋势预测,然后根据季节指数的趋势和季节指数与季节指数趋势之间的比值之乘积,得到修正的季节指数,以提高季节指数中含有的季节特征信息。实验结果显示,这种预测模型在现有实验数据环境中最大相对误差为3.08%,远低于原交乘趋向模型的8.42%。Cross Multiply Trend Model is one of Season Forecasting Method, it can increase the season swing according to historical data trend, but the feature of season contained in historical data can not be extracted adequately when get season index by average historical season index simply. This paper describes a new model,, season index is modified by modify factor in order to enrichment the feature information of season index. By experiments, the results show that the max relative forecasting error of new model is 3.08 %, more below than old model＇ s 8.42%.

关 键 词：季节模型 季节指数 交乘趋向模型 灰色系统 时间序列 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]