Font Size: a A A

The Application Of Quantile Regression Based On The Correction Multicollinearity Method

Posted on:2018-08-05Degree:MasterType:Thesis
Country:ChinaCandidate:X KaiFull Text:PDF
GTID:2359330536455613Subject:Probability theory and mathematical statistics
Abstract/Summary:PDF Full Text Request
Quantile regression is not limited to the distribution of random errors,it can deal with spikes and thick tail data effectively.However,use the quantile regression in the data directly,often affected by the multiple collinearity between the independents.It is necessary to eliminate the collinearity between the independents before the quantile regression.This paper introduces the mathematical idea and the basic nature of the of quantile regression,and takes the melon production data of the Xinjiang as an example.The mathematical model of the quantile regression is established.And then through the significance test and Wald test results,found that the establishment of the model is ineffective.Thus,for the covariance of independent variables,the reason for the nature of the model is influenced,and the idea of ??combining the multiple collinearity correction method with the quantile regression is produced.Then,the common methods of multi-collinearity correction: principal component regression,ridge regression and partial least squares regression are combined with quantile regression respectively.Three kinds of binding methods are applied to the data.Finally,by comparing the advantages and disadvantages of the three methods,we choose the relatively superior combination method,analyze the influence factors of the yield of melon in Xinjiang.Structure of this paper: In the first chapter,we describe the background,the significance of the research,the research status,the research content and the research method.In the second chapter,we discuss the regression model of quantile regression.In the third chapter,we discuss the multi-collinearity method combination of quantile regression,including the quantile regression under the principal component method,the quantile regression under the ridge regression method,the quantile regression under partial least squares regression;Chapter 4,comparing the three methods of combining,combining the advantages and disadvantages of the actual analysis;Chapter 5,General,summarizes the summary of the methods in the article,and puts forward the shortcomings and prospects.
Keywords/Search Tags:Quantile Regression, Multiple Collinearity, Principal Component Regression, Ridge Regression, Partial Least Square Regression
PDF Full Text Request
Related items