Font Size: a A A

Research On Interval Combination Prediction Model Under Several Kinds Of Prediction Indicators

Posted on:2018-11-22Degree:MasterType:Thesis
Country:ChinaCandidate:C J W WeiFull Text:PDF
GTID:2359330542463709Subject:Quantitative Economics
Abstract/Summary:PDF Full Text Request
Prediction is based on the development and management of social economy and has been developed and developed for thousands of years.Scientific prediction is the foundation of scientific decision making,which makes people realize the importance of prediction to decision making,so it is predicted that this subject will be produced.Because there are a lot of uncertainty in the social life,these uncertainties to make people feel more to understand and grasp the future development necessary and urgent things,this is the power of predicting the subject development.Due to various factors,the prediction object is not necessarily single,it can be a complex system.So the use of traditional single forecasting model to forecast there are defective,in order to improve the prediction precision of combination forecast method gradually appear in people's field of vision,and extensive concern and attention.Because the objective world is complex and the human mind is imprecise,people often encounter many uncertain or incomplete information,Coupled with the rapid development of social and economic,scientific and technological progress and the abundance of the knowledge of information made in many fields such as economy,management,military,etc.,in some decision problems expressed in the form of interval number and fuzzy number.Therefore,the study of interval Numbers is of great theoretical significance and practical background.This paper focuses on the study of interval number combination prediction from the non-optimal and the optimal,the right and the right,the distance measure index and the similarity measure index.This article selects a distance measure,a similarity measure,and the research of information entropy index,from the three aspects are to build the interval combination forecasting model based on the different indicators.In this paper,we construct five prediction models of interval combination,and verify the validity of the model from two aspects of theory and case.The main contents of this paper can be divided into the following three aspects:(1)Based on information entropy,the model of the combination of non-optimal positive weights is established.Based on the sequence of the interval number,the method of m species is used to predict the sequence,and the corresponding interval number is calculated.The interval information entropy formula is constructed,and the method of calculating the weight of the interval combination prediction model is given in this Angle,and the validity of the model is verified by the data instance operation of interval data.Although the model is not necessarily an optimal portfolio prediction model,it provides a simple method to calculate the weight coefficient of the interval portfolio prediction model.(2)Introducing Hausdorff distance interval number vector,and as a rule of target based on Hausdorff distance weighting interval is constructed based on GIOWA operator combination forecast model and variable weight combination forecasting model of Hausdorff distance.Firstly,the validity of the model is proved by the two aspects of superiority and redundancy,and the validity of the model is verified by the data.Combination forecast model for variable weight range,take the GIOWA operator parameters when three conditions to verify its validity,finally,the sensitivity analysis of parameters for the given the change of each single prediction methods,error evaluation index weights and the influence of the objective function value.(3)Building combining norm similarity and similarity in the direction as the target,a new vector similarity criterion,based on COWA operator weighting interval combination forecasting model with variable weight interval combination forecasting model based on ICOWA operator.The validity of the two models is verified by the selection of data,and the sensitivity analysis is made for the attitude parameter values.In this paper,the change of the weight of the three methods in the combination prediction is analyzed,and the influence of the change on the evaluation index and the ultimate objective function value is also discussed.In the end,the thesis generally generalizes the article,points out the innovation and inadequacy of the article,and makes a forecast for the more in-depth study.
Keywords/Search Tags:Interval combination forecasting, Entropy value method, Hausdorffdistance, Vectorsimilarity, GIOWA operator
PDF Full Text Request
Related items