摘要
单一的预测模型在对时间序列数据进行预测时会产生预测精度差和适用性差的问题。文章首先利用回归分析、三次指数平滑法、ARIMA模型、GM(1,1)模型等单一模型对区域天然气需求进行预测;然后采用加权组合预测思想,运用模糊综合评价法(FAHP法)对各个模型进行评分,剔除评分最低的预测模型以避免影响最后的加权组合精度;最后将FAHP法的评分域变换到加权组合模型的权数域中,对区域天然气需求进行加权组合预测。实证结果表明:与单一预测模型和传统的加权组合模型相比较,基于FAHP评分及权重确定的加权组合模型可将预测精度最高提升9.47%,明显高于单一预测模型和传统加权组合模型。
Single forecasting model will cause poor accuracy and weak applicability when predicting time series data. This paper firstly uses the single model--regression analysis, three exponential smoothing method, ARIMA model and GM (1,1) model- to predict the demand of natural gas in a region. And then the paper adopts weighted array idea and fuzzy analysis hierarchy process (FAHP) to grade each model and reject the one with the lowest score so as to avoid its negative influence on the accuracy of the final weighted array. Finally, the paper transfers the score field of the FAHP to the weight field of the combined model and forecasts the regional gas demand with the weighted array model. The result demonstrates that the weighted array model based on FAHP scoring and weight is able to increase the prediction accuracy to the highest 9.47%, obviously higher than single model and traditional weighted array models.
出处
《统计与决策》
CSSCI
北大核心
2017年第23期74-78,共5页
Statistics & Decision
基金
国家自然科学基金资助项目(71402094)
教育部人文社会科学青年基金资助项目(13YJC630210)
上海海事大学校基金资助项目(20120080)
关键词
模糊综合评价法
组合模型
需求预测
fuzzy analysis hierarchy process
combination model
demand forecasting