Multiple Hypothesis Testing Method And Applications

In the analysis of high dimensional data,the application of hypothesis testing tends to be more and more refined.Because the traditional single-weight hypothesis test is to take separate test of each hypothesis,and there is no previous consideration of other hypothetical tests,which are tested as a whole.Therefore,from the overall point of view of the test,at this time the probability of making the first type of error will be much greater than the previously selected?,and thus leads to some wrong conclusions.In order to solve the above problems,the problem of multiple hypothesis tests that appeared in the the 1950 s has attract statisticians’ attention and has been greatly developed.In the existing domestic literature,the multiple hypothesis test method is seldom applied to China’s economic data analysis.Therefore,based on the multiple hypothesis testing method,this paper analyzes the application of income gap of Chinese residents.The first chapter summarizes the background and research significance of the problems studied in this paper,the literature review and the main work to be done in this paper.In the second chapter,the basic concept of hypothesis test and the definition of multiple hypothesis test are reviewed,and several error measurement methods and related control algorithms of multiple hypothesis test are discussed.The third chapter is a statistical analysis of the description of changes in income inequality in China.In the fourthand fifth chapters,based on the application of false discovery rate(FDR)control algorithm,through the empirical analysis of two datasets,the variables that have a significant impact on the income difference of residents are selected.In the sixth chapter,according to the results of empirical analysis and the empirical results of introducing others,the concrete application of multiple hypothesis tests in practical problems is obtained by comparative analysis:(1)the more variables there are,the better the multiple hypothesis test effect;(2)the more the p-value distribution of variables conforms to the U(0,1)distribution,the better the multiple hypothesis test effect.In the last chapter,the main contents of this paper are summarized,and the future research direction is prospected.