對于有自我保護現(xiàn)象的隨機化回答數(shù)據(jù)的半?yún)?shù)貝葉斯回歸分析.pdf

1、目錄艘1Abstract2一、弓I言3二、模型及先驗82.1.樣本數(shù)據(jù)及其模型82.2先驗分布92.3.Dirichlet過程簡介11三、后驗分布的推導及模擬方法143.1.模型的貝葉斯估計方法143.1.1引入潛變量名和谷143.1.2引入潛變量^173.1.3g的滿條件分布183.2.Gibbs抽樣方法203.3.統(tǒng)計推斷21四、模擬研究244.1方法的使用244.2.模型的比較29五、實例分析36參考文獻：39AbstractRo

2、mizedResponseisacommonsurveymethodintheinvestigationofsensitiveissuesBecauseoftheintentiontoprotectselfprivacyininvestigationrespondentsstillexhibitselfprotectivebehaviinRomizedResponseSystem.Thispaperanalyzesthestatisti

3、calmethodofromizedresponsedatainthepresenceofselfprotectivebehavi.WeuseDirichletprocessasthepridistributionoflinkfunctions.Andthenwecanobtainthemarginalpostdistribution(fullconditionaldistribution)givetheGibbssamplingmet

4、hod.Thesimulationresultsshowsthatinthemediansamplevolumeourmethodisbetterthantheparametricmethodwhenthelinkfunctionofparametricmethodisdesignedwrong.Finallyweuseourmethodinanexamplecomparedourmethodwiththeparametricmetho