回归模型的估计方法与在林业中的应用分析

《回归模型的估计方法与在林业中的应用分析》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、thedifferentobjectandobservatingmethodofobservationdatainforestry,theheteroscedasticityisunavoidableanddiffersindegree．Basingonformerstudy,theresearchonheteroscedasticityinheight—diametermultiplelinearmodelsofP加uskesiyavar．1angbianensisismadewithtwelvekindsofweightingfunct

2、ionandfourkindsofvariable．weightingfunction．Atsametime，theheteroscedasticiytoone．wayvolumemodeloflarchisstudiedbyeightkindsofweightfunction．Theweightingfunction1压“”ischosentoeliminateheteroscedasticityofheight．diametermultiplelinearmodelsofPinuskesiyavar．1angbianensis，whil

3、eweightingfunction1，D‘Hischosentoeliminateheteroscedasticityofone．wayvolumemodeloflarch．Themethodofweightedregressionestimateeliminatingheteroscedasticityprovesthatvariousmodelhasdifferentoptimalweightfunctionform．Wheneliminatingheteroscedasticity,variousweightfunctionform

4、sshouldbetestedinordertoselecttIlebestone．Theweightfunctioniscannotbeconstructediustbyusualweightfunctionform，althoughheteroscedasticityinmodelcanalsobeeliminated．resultisunrefinedandnotbest．Forsakeofprecision，variable．weightingfunctionshouldbeusedinselectingofweightfuncti

5、on。绛西enheteroscedasticityshowslinearstructure，thervalueofvariable．weightingfunctionisrelatedtointensityofhetemscedasticityinmodeLmatissamemodelanddifferentheteroscedasticityhasvarious厂．Thestrongerofheteroscedasticityinmodel．themoredeviationtozeroofrvalue．oppositely,thenear

6、ertozeroofrvalueandnearert01ofweiglltfunction，whichlooksmorelikeleastsquaresregression．Whenheteroscedasticityshowsdecreaseinlinearthatislinearslopeofvarianceisnegative，厂ispositive．Theslowerofslopedecreaed，thelargerof厂value，andviceversa．3．TheeffectofnleasureInente1170ronvol

7、unlemodelV=aD6isstudiedinchapterfive，and。theparameterestimatingismadebymeasurementerrors，whichdrawfollowingconclusions：Withincreasingofdeviationofdependentvariablewithmeasurementerror,theestimatingparametersfluctuatesandbecomesunsteadyintrend，whenmeasureerrorexistsindepend

8、entvariablebutconstrarilyinindependentvariable．Themeasurementerrorofdependentvariableaffe