Fuzzy Variable Weight Combination Forecasting Methods

This paper mainly studied variable weight combination forecast, improved the method of fuzzy variable weight combination(FVW), brought forward a method of forecasting efficiency evaluation ,which not only calculated the mean errors of a forecasting model ,but also analyzed the reasons of errors thoroughly.The paper is composed of three chapters. The first one, reference summarization, introduced the basic definitions of combination forecast, the present developing state and exiting problems of combination forecast, the late theory achievements of combination forecast. The second one .the study on fuzzy variable combination forecast, did a few trial work in three aspects on the basis of references outside and inside. Firstly, this paper put forward a wholly and systematic model-selection system from the selection of single forecasting model to the selection of group forecasting models which took part in the combination forecast; secondly, this paper improved the method of fuzzy variable combination in paper [8]; thirdly, with the help of probability theory and grey system theory, put forward a wholly, systematic, convenient and feasible evaluation system of combination forecast, the method of grey incidence error analysis of which ,analyzing every reasons which led to the errors of combination forecast ,is pioneering work in the field of forecasting model evaluation. The third one, application, applied the improved method of FVW to the forecast of the sale sum of goods each month in a city, compared it with the method in paper [8] and the methods which participated in the combination forecast .The result showed that the method in this paper fit much more to the reality and reflected the varying rules of things more .It is successful in a way.