Research And Application Of Combined Forecasting Model

With the rapid development of social economy and the continuous progress of science and technology,the forecasting method is playing an irreplaceable role.In short,it is predicted that the known information is expected to happen in the future or impossible to happen,which is essentially a process of the necessary cognitive and rational analysis of the development and change of things.So far,the types of forecasting methods have reached hundreds,each model is from different angles to extract useful information,they have their own advantages and disadvantages and adaptation conditions.In the prediction practice,if the multiple single prediction methods are used together correctly,the prediction results will be less sensitive to a single poor prediction method,so as to improve the accuracy and reliability of the forecast.In 1969,Bates and Granger first proposed the idea of combined forecasting,that is,the combination of various individual forecasting models in appropriate weighting to obtain the final predictions.In the whole prediction model,how to solve the weighted average coefficient is the most important.In this paper,the optimal combined forecasting model based on multiple criteria is established,and the method of determining the weight coefficient of multiple combinations is put forward,and the application research of combined forecasting in complex socioeconomics is discussed.Through the multidimensional study,we can find that the existing traditional combined forecasting model has some shortcomings.It has been unable to meet the social needs.It is based on the method of single prediction to give them different weighted average coefficient,the same single prediction method in the sample interval at each time point of the weighted average coefficient is constant,it is clear that this does not meet the real requirements.In order to solve this problem,this paper proposes a combined forecasting model based on the induced orderly weighted harmonic averaging(IOWHA)operator.This article first investigates the background and significance of the subject,understands the problem and finds the solution.Secondly,we understand the relevant forecasting technology in detail,and systematically learn and master the way to determine the weight of the combined forecasting model.Then,the linear and nonlinear combined forecasting model based on the error index is expounded,and a number of suitable practical cases are selected to carry out the simulation experiment respectively.And the Artificial Bee Colony Algorithm is introduced to determine the weight of the optimal combination forecasting model.The purpose is to solve the problem that the workload is large and the guaranteed weight is equal to zero.Finally,by introducing the IOWHA operator,the problem of the traditional optimal variable weight combined forecasting model is well solved.Based on this,two kinds of concepts of the second-order predictive validity and the geometric distance based on the LI norm are expounded,and two new models are combined with IOWHA operator,and their research and description are respectively.Finally,combined with actual cases,we made a detailed analysis.Through a number of examples,it is shown that the combined forecasting model is superior to the single prediction model in terms of prediction effect and overall performance.Moreover,the prediction results of the two improved IOWHA combined forecasting models proposed in this paper are the most outstanding,which not only overcome the shortcomings of the traditional combination forecasting model,but also greatly improve the prediction accuracy and prediction accuracy,having some validity and can be applied to reality.