帶相依結(jié)構(gòu)的二維列聯(lián)表的多重比較
發(fā)布時(shí)間:2021-07-09 20:38
列聯(lián)表是一種常見的數(shù)據(jù)存儲(chǔ)格式,其中的數(shù)據(jù)是將觀測(cè)數(shù)據(jù)按兩個(gè)或者更多屬性進(jìn)行分類后的頻數(shù)。列聯(lián)表常用于醫(yī)學(xué)、生物學(xué)、社會(huì)科學(xué)等學(xué)科之中。通過對(duì)列聯(lián)表進(jìn)行統(tǒng)計(jì)分析,可以考察各個(gè)屬性之間是否有聯(lián)系,也就是判斷兩個(gè)屬性變量是否具有獨(dú)立性.可用于檢驗(yàn)列聯(lián)表中的兩個(gè)屬性變量的獨(dú)立性的假設(shè)檢驗(yàn)方法有:卡方獨(dú)立檢驗(yàn)、Fisher精確檢驗(yàn)、Cochran-Mantel-Haenszel檢驗(yàn)[1]等。本文的目的是建立一個(gè)模型,進(jìn)而基于這個(gè)模型提出一個(gè)新的檢驗(yàn)方法,用于檢驗(yàn)列聯(lián)表的屬性變量的獨(dú)立性,這個(gè)模型也可以衡量兩個(gè)屬性變量的(正、負(fù))相關(guān)性。本文的整體思路是通過引入一個(gè)可以衡量變量相關(guān)性的參數(shù),構(gòu)造一個(gè)二元二項(xiàng)分布,并將它用于對(duì)二維列聯(lián)表的統(tǒng)計(jì)分析。在緒論中,我們總結(jié)了文獻(xiàn)中已有的二元二項(xiàng)分布。假設(shè)隨機(jī)向量(X,Y)服從二元二項(xiàng)分布,且兩個(gè)邊際分布都是二項(xiàng)分布,即X~Binomial(n1,π2)和Y~Binomial(n2,π2)。文獻(xiàn)中已有的二元二項(xiàng)分布大致可以分成三類:第一種是要求n1與n2相等,即n1=n2=n;第二種是要求π1和π2相等,即π1=π2=π;第三種對(duì)n1,n2和π1,π2不做...
【文章來源】:哈爾濱工業(yè)大學(xué)黑龍江省 211工程院校 985工程院校
【文章頁數(shù)】:70 頁
【學(xué)位級(jí)別】:碩士
【文章目錄】:
摘要
Abstract
Chapter 1 Introduction
1.1 Background and Significance
1.2 Literature Review
1.3 Thesis Outline
Chapter 2 A New Bivariate Binomial Model
2.1 Introduction of Model
2.2 Theoretical Properties
2.3 Estimation of Parameters
2.3.1 Maximum Likelihood Estimation
2.3.2 Bootstrap Confidence Interval
2.3.3 Parameter Estimation of Incomplete 2 × 2 Contingency Tables
2.4 Test of independence
2.4.1 Likelihood Ratio Test
2.4.2 Score Test
2.4.3 Wald test
2.4.4 Bootstrap Hypothesis Test
2.4.5 Bartlett-Corrected LRT
2.5 Brief Summary
Chapter 3 Multiple Comparison of Dependency Parameter
3.1 Multiple Comparison Problem
3.2 Controlling Methods for p-value
3.2.1 Bonferroni's Method
3.2.2 Sidak's Method
3.2.3 Holm's Method and Hochberg's Method
3.2.4 Hommel's Method
3.2.5 Benjamini-Hochberg Method and Benjamini-Yekutieli Method
3.3 Resampling Method
3.4 Brief Summary
Chapter 4 Simulation Studies
4.1 Accuracy of Point Estimators and Interval Estimators
4.2 Performance of Three Asymptotic Tests and the Bootstrap Test
4.3 A Real Data Example
4.4 Multiple Comparison
4.5 Brief Summary
Conclusions
結(jié)論
References
Appendix A First and Second Partial Derivatives of (2-23)
Appendix B Cumulants and their Derivatives for the Bartlett Correction Factor
Acknowledgements
本文編號(hào):3274456
【文章來源】:哈爾濱工業(yè)大學(xué)黑龍江省 211工程院校 985工程院校
【文章頁數(shù)】:70 頁
【學(xué)位級(jí)別】:碩士
【文章目錄】:
摘要
Abstract
Chapter 1 Introduction
1.1 Background and Significance
1.2 Literature Review
1.3 Thesis Outline
Chapter 2 A New Bivariate Binomial Model
2.1 Introduction of Model
2.2 Theoretical Properties
2.3 Estimation of Parameters
2.3.1 Maximum Likelihood Estimation
2.3.2 Bootstrap Confidence Interval
2.3.3 Parameter Estimation of Incomplete 2 × 2 Contingency Tables
2.4 Test of independence
2.4.1 Likelihood Ratio Test
2.4.2 Score Test
2.4.3 Wald test
2.4.4 Bootstrap Hypothesis Test
2.4.5 Bartlett-Corrected LRT
2.5 Brief Summary
Chapter 3 Multiple Comparison of Dependency Parameter
3.1 Multiple Comparison Problem
3.2 Controlling Methods for p-value
3.2.1 Bonferroni's Method
3.2.2 Sidak's Method
3.2.3 Holm's Method and Hochberg's Method
3.2.4 Hommel's Method
3.2.5 Benjamini-Hochberg Method and Benjamini-Yekutieli Method
3.3 Resampling Method
3.4 Brief Summary
Chapter 4 Simulation Studies
4.1 Accuracy of Point Estimators and Interval Estimators
4.2 Performance of Three Asymptotic Tests and the Bootstrap Test
4.3 A Real Data Example
4.4 Multiple Comparison
4.5 Brief Summary
Conclusions
結(jié)論
References
Appendix A First and Second Partial Derivatives of (2-23)
Appendix B Cumulants and their Derivatives for the Bartlett Correction Factor
Acknowledgements
本文編號(hào):3274456
本文鏈接:http://sikaile.net/kejilunwen/yysx/3274456.html
最近更新
教材專著