基于IGOWFPA算子的组合预测模型

The Combined Forecasting Model Based on IGOWFPA Operator

集成算子是当前组合预测研究的一个热点,早期的关于算数平均算子、几何平均算子、调和平均算子的理论已经成熟。但是目前关于广义型集成算子的研究还比较少,很多地方值得更深入细致地研究。首先探讨了什么样的函数可以与加权比例平均算子结合(即单调函数),将现有与"加权比例平均算子"结合的"指数函数、对数函数"推广为一般的单调函数;其次,提出加权函数比例平均(WFPA)系列算子,研究了这些算子的性质;最后,建立了基于IGOWFPA算子的新的组合预测模型,以5类常见单调函数为例进行了实证分析,结果显示误差较小,且具一般性。

The integration operator is a hot spot in the current research of combined forecasting.The early theories about arithmetic average operator,geometric mean operator and harmonic mean operator have already entered a mature stage.However,there are few researches on generalized integration operators at present,which deserves more detailed research.Firstly,this paper explores what kind of function can be combined with the weighted proportional averaging operator(i.e.,monotone function)and generalizes the exponential function and logarithmic function combined with the weighted proportional averaging operator to a general monotone function.Secondly,the weighted function proportional average(WFPA)series operator is put forward.Finally,a new combined forecasting model based on IGOWFPA operator is established.An empirical analysis of five kinds of common monotone functions is carried out.The results show that the error is smaller and more general.