Optimal Interval Combination Forecasting Method And Its Application

The rapid development of the world economy and science and technology is accompanied by more and more uncertain factors.The existence of these factors makes people gradually realize the importance and urgency of understanding and mastering the future.The prediction can provide a more science theoretical reference for the society and the management department,and this reality makes it possible to predict the continuous development of this discipline.In the actual forecasting process,the complexity of the forecasting object makes the accuracy of the traditional single forecasting method biased.If discarded,some important information may be lost,resulting in a decrease in forecasting accuracy.The combined forecasting method just solves this defect and has received extensive attention and application.With the development of the objective world,the fuzziness of people's thinking and uncertainty of environment make the prediction information not a definite value.This paper will turn the study of real numbers into the sequence of interval numbers,and predict the economic variables that appear in the form of intervals,Studying it has better practical reference value.In order to enrich the method of interval number combination forecasting,this paper starts with the conversion of interval numbers into real numbers?triangular fuzzy numbers and ternary contact numbers.Six interval-type combined forecasting models are constructed,and their validity and the advantages and disadvantages of method are proved theoretically and example verification,the specific content is divided into the following three parts:(1)The interval number is transformed into a real number.The interval combination forecasting model based on Jaccard similarity coefficient is constructed according to the concept and formula of Jaccard similarity coefficient.For the Jaccard's similarity coefficients,the weighted interval combination forecasting model is given,the definition of its validity and advantages and disadvantages are given.The data is used to verify the model constructed.Finally consider the specificity of the Sensitivity analysis of the influence of different values,The sensitivity analysis of the influence of different values of parameter between 0 and 1 on the weight value.For the variable weight interval combination forecasting model of Jaccard's similarity coefficient,the relevant advantages and disadvantages of the model are defined.Three special values of the generalized ICOWA operator are selected to verify the advantages and disadvantages and effectiveness of the model.(2)The number of intervals is expressed in the form of triangular fuzzy numbers.The similarity of triangular fuzzy numbers is introduced as a new optimal criterion,and an interval combined forecasting model based on the similarity of triangular fuzzy numbers is constructed.For the weighted interval combination forecast model of similarity of triangular fuzzy numbers,theoretical analysis was performed on the advantages and disadvantages and redundancy of the model.Finally,the data was selected to verify the model.For the variable weight interval combined forecasting model of similarity of triangular fuzzy numbers,the introduced GIOWA operator selects three special values,and the model is verified by examples.Finally,the sensitivity analysis is carried out for different cases of parameter,and the influence of on the value of weight?the value of objective function and the system of error index value between-2 and 2.(3)The interval number is transformed into the form of three-element contact number,and the interval combination forecasting model based on the similarity degree of contact numbers is constructed according to the related concept of the similarity of contact numbers.A weighted interval combination forecasting model for the similarity of contact numbers was used to carry out a detailed theoretical analysis,and then the data was selected for further case validation.For the variable weight interval combined forecasting model of the contact number similarity,three special values in the GIOWA operator are selected to verify the variable weight model.Finally,further research and analysis are carried out on the influence of parameter on the weight value,objective function value and error index value of the parameter in the interval 0 to 2.The last chapter summarizes the contents of the full text,points out the deficiencies of the article,and makes inferences about where this article requires more in-depth research.