会计学（一）53例接受手术的前列腺癌患者(huànzhě)情况（二）26例冠心病病人和28例对照进行(jìnxíng)病例对照研究26例冠心病病人和28例对照者进行(jìnxíng)病例对照研究一、logistic回归(huíguī)模型概率(gàilǜ)预报模型二、模型(móxíng)的参数估计参数估计的公式(gōngshì)三、回归(huíguī)参数的假设检验优势(yōushì)比及其可信区间标准化回归(huíguī)参数SAS程序(chéngxù)TheLOGISTICProcedureAnalysisofMaximumLikelihoodEstimates预报(yùbào)模型TheLOGISTICProcedureAnalysisofMaximumLikelihoodEstimates预报(yùbào)模型四、回归(huíguī)参数的意义优势(yōushì)比改变exp(bj)个单位/五、整个回归(huíguī)模型的假设检验似然比检验(jiǎnyàn)（likelihoodratiotest）ROC曲线(qūxiàn)模型评价ROC曲线(qūxiàn)模型评价六、logistic逐步回归（变量(biànliàng)筛选）模型(móxíng)中有X5、X6、X8，看是否引入X1第二节.条件(tiáojiàn)logistic回归条件(tiáojiàn)似然函数1:3配对(pèiduì)的例子1:2配对(pèiduì)的例子表16-7条件(tiáojiàn)logistic回归的SAS程序结果(jiēguǒ)西、中西、中三种(sānzhǒnɡ)疗法哑变量化SPSS软件(ruǎnjiàn)计算DATAsamp16_1;INPUTx_raygradestageageacidnodes;CARDS;......;PROCLOGISTICDESCENDING;MODELnodes=x_raygradestageageacid/RISKLIMITS;OUTPUTOUT=predPROB=pred;PROCPRINTDATA=pred;RUN;TheSASSystem22:07Monday,November29,20051TheLOGISTICProcedureModelInformationDataSetWORK.SAMP16_1ResponseVariablenodesNumberofResponseLevels2NumberofObservations53ModelbinarylogitOptimizationTechniqueFisher'sscoringResponseProfileOrderedTotalValuenodesFrequency11202033Probabilitymodeledisnodes=1.ModelConvergenceStatusConvergencecriterion(GCONV=1E-8)satisfied.ModelFitStatisticsInterceptInterceptandCriterionOnlyCovariatesAIC72.25260.126SC74.22271.948-2LogL70.25248.126TestingGlobalNullHypothesis:BETA=0TestChi-SquareDFPr>ChiSqLikelihoodRatio22.126450.0005Score19.451450.0016Wald13.140650.0221TheSASSystem22:07Monday,November29,20052TheLOGISTICProcedureAnalysisofMaximumLikelihoodEstimatesStandardWaldParameterDFEstimateErrorChi-SquarePr>ChiSqIntercept10.06183.45990.00030.9857x_ray12.04530.80726.42080.0113grade10.76140.77080.97590.3232stage11.56410.77404.08350.0433age1-0.06930.05791.43200.2314acid10.02430.01323.42300.0643TheSASSystem22:07Monday,November29,20052TheLOGISTICProcedureAnalysisofMaximumLikelihoodEstimatesStandardWald