異方差與序列相關(guān)性練習(xí)

1、一、異方差檢驗與修正一、異方差檢驗與修正（一）建立初始回歸模型（一）建立初始回歸模型相關(guān)命令相關(guān)命令：datadataxyscatscatxylslsycx模型一：DependentVariable:YMethod:LeastSquaresDate:102314Time:10:46Sample:120Includedobservations:20VariableCoefficientStd.ErrtStatisticProb.C272.

2、3635159.67731.7057130.1053X0.7551250.02331632.386900.0000Rsquared0.983129Meandependentvar5199.515AdjustedRsquared0.982192S.D.dependentvar1625.275S.E.ofregression216.8900Akaikeinfocriterion13.69130Sumsquaredresid846743.0S

3、chwarzcriterion13.79087Loglikelihood134.9130Fstatistic1048.912DurbinWatsonstat1.301684Prob(Fstatistic)0.000000（二）異方差的四種檢驗方法及其分析（二）異方差的四種檢驗方法及其分析右擊右擊residresid選擇選擇ObjectObjectCopyCopy，輸入，輸入e得到初始回歸模型的殘得到初始回歸模型的殘差序列；差序列；1.1

4、.圖示法：圖示法：scatscatxe^2e^2Sample:18Includedobservations:8VariableCoefficientStd.ErrtStatisticProb.C1277.1611540.6040.8290000.4388X0.5541260.3114321.7792870.1255Rsquared0.345397Meandependentvar4016.814AdjustedRsquared0.2362

5、96S.D.dependentvar166.1712S.E.ofregression145.2172Akaikeinfocriterion13.00666Sumsquaredresid126528.3Schwarzcriterion13.02652Loglikelihood50.02663Fstatistic3.165861DurbinWatsonstat3.004532Prob(Fstatistic)0.125501子樣本模型二：De

6、pendentVariable:YMethod:LeastSquaresDate:102314Time:10:57Sample:1320Includedobservations:8VariableCoefficientStd.ErrtStatisticProb.C212.2118530.88920.3997290.7032X0.7618930.06034812.625050.0000Rsquared0.963723Meandepende

7、ntvar6760.477AdjustedRsquared0.957676S.D.dependentvar1556.814S.E.ofregression320.2790Akaikeinfocriterion14.58858Sumsquaredresid615472.0Schwarzcriterion14.60844Loglikelihood56.35432Fstatistic159.3919DurbinWatsonstat1.7229

8、60Prob(Fstatistic)0.000015根據(jù)得到的根據(jù)得到的RSSRSS1與RSSRSS2，求得，求得F檢驗統(tǒng)計量值。檢驗統(tǒng)計量值。F=F=RSSRSS2RSSRSS1=615472.0126528.3=4.86=615472.0126528.3=4.86查F分布表，確定臨界值分布表，確定臨界值F0.050.05（6666）；）；若FFF0.050.05（6666）則拒絕）則拒絕H0，認為原初始模型的隨機誤差，認為原初始模型