复杂数据下约束线性模子的统计推断.pdfVIP

ABSTRACT Inthis first the andmain of paper,wegeneralizedevelopmentproperty linear modelintroducestatisticalinferenceforthe regressionand linear modelwhose coefficientswithconstraints．Basedon regression regression these themeasurementerror in theories，we model(error study independent variableandthe variablecanbe variable dependent accuratelyobserved)and with values valuesin variableand missing model(missing dependent variablecanbe weintroducethe independent observed)．Then completely ofthesetwo the estimationforthese development models，giveparameter two data derivetheconstrainedleast estimation models，and complex squares for tobeestimatedwith the parameters constraints．ByLagrange multiplier constructtheteststatisticbasedonthedifference method，we ofresidualsum of betweennull andalternative ofconstraints squares hypothesis hypothesis andWaldstatistictotestthe concludethatthe constraints，we teststatisticwe haveconstructedfollowthe distribution whennull chi—square