The Improvement Of Two Kinds Of Fuzzy Regression Model And The Application Research

Fuzzy regression analysis have a wide range of applications in engineering andtechnical, economic, financial, management, biological science and other fields, the studyof fuzzy regression model become the focus of many scholars’ research naturally, fuzzyregression analysis model and estimates of regression coefficients are two importantaspects. In the previous study, most people use symmetric triangular fuzzy number as thecoefficient of the fuzzy linear regression model, which are often neglected other types offuzzy number.First, introducing the development process of the fuzzy mathematics and fuzzyregression analysis, presents several fuzzy number and fuzzy number distance definition;and by changing the weight function expression f (λ), get a distance definition ofGaussion fuzzy number.Secondly, the parameter estimation of a fuzzy regression model based on theGaussion fuzzy number is introduced, using some statistics of the classical regressionanalysis to the evaluation of the Gaussion fuzzy number regression model, and analyzingthe fitting performance of this method through an example.Then, a fuzzy varying coefficient regression model and its parameter estimations werediscussed. On one hand, analyzing the model for reducing the influence of abnormal valuewhen estimating the model parameters, on the other hand, using numerically simulated forcomparing our method with the existing methods. Finally, the fuzzy varying coefficient robust regression model is applied to theanalysis of GDP and its influence factors, then forecasting the GDP of2009, the resultsshowing that the model has better prediction ability.