回归分析Oracle性质,几类回归模型中的变量选择方法.pdf

进行数值模拟，验证得出ADS比lasso等变量选择方法更具适用性和可行

性．

半参数模型

Ⅱ

VARIABLESELECTIoNMoTHoDINSEVERAI。

REGRESSIONMODELS

ABSTRACT

of也evariableselectionmethodiStochoosethe

Themainfunction

SOastoestablisha

usefulvariabletoreducethedimension simple

appropriate

model．The itsrelatedmethodsarenew

Dantzigselector(DS)and high

dimension

variableselectionmethodfordatas眦恤which

dimensionaldata

selectionand

is thanthatof canachievethevariable parameter

greater sample．It

estimation this thecharacterizationand

thesis，we applications

together．In study

severalstatistical

oftheDSand selector(ADS)in models，the

adaptivedantzig

areasfollows：

mainresearchwehavecarriedonandtheirresults

variable

tothecommon forward

variable，weput

(1)According counting

selectionand estimationforthePoisson modelvia

linearity

parametric logarithm

underthe

onthe ofADS

theDSand focus asymptoticproperty

ADS．Emphasize

Poisson model．Provethe ofnonzerocoefficient