某地區(qū)某農(nóng)產(chǎn)品收購(gòu)量與

某地區(qū)某農(nóng)產(chǎn)品收購(gòu)量與年份YX196935.5819.6919557.965.33197031.0321.13195615.346.82197114.3332.34195713.558.17197213.8819.57195810.949.48197314.6616.3919596.398.03197419.3717.9619601.943.58197535.4718.3919610.061.17197635.4918.8319620.660.92197732.7721.1519636.041.63197832.219.8196415.417.73197938.534.86196515.39.46198053.7222.96196619.3213.97198151.317.45196735.7617.32198234.0416.05196835.0317.36198316.0317.38198421.7916.79第一步：描述統(tǒng)計(jì)量YX

Mean

22.35857

14.55500

Median

17.36500

16.85500

Maximum

53.72000

34.86000

Minimum

0.060000

0.920000

Std.Dev.

14.95748

8.646472

Skewness

0.324925

0.293930

Kurtosis

2.144772

2.808145

Jarque-Bera

1.346006

0.446120

Probability

0.510174

0.800067

Sum

626.0400

407.5400

SumSq.Dev.

6040.606

2018.560

Observations

28

28由表可知：y和x的最值，平均值，中位數(shù)，方差等數(shù)據(jù)。由經(jīng)濟(jì)理論知，農(nóng)產(chǎn)品收購(gòu)量受銷(xiāo)售量影響，當(dāng)銷(xiāo)售量增加時(shí)，農(nóng)產(chǎn)品收購(gòu)量也會(huì)增加，它們之間有同步變動(dòng)趨勢(shì)，農(nóng)產(chǎn)品收購(gòu)量除受銷(xiāo)售量的影響之外，還受到其他一些變量及隨機(jī)因素的影響，將這些變量均歸并到隨機(jī)變量u中，根據(jù)X與Y的樣本數(shù)據(jù)作散點(diǎn)圖散點(diǎn)圖由圖可知，它們的變化趨勢(shì)是線(xiàn)性的，由此建立農(nóng)產(chǎn)品收購(gòu)量與銷(xiāo)售量之間的一元線(xiàn)性回歸模型，通過(guò)Eviews一元線(xiàn)性回歸模型參數(shù)DependentVariable:YMethod:LeastSquaresDate:06/10/12Time:16:31Sample:19551982Includedobservations:28VariableCoefficientStd.Errort-StatisticProb.

C4.8570824.1098471.1818160.2480X1.2024380.2439014.9300350.0000R-squared0.483155

Meandependentvar22.35857AdjustedR-squared0.463277

S.D.dependentvar14.95748S.E.ofregression10.95806

Akaikeinfocriterion7.694777Sumsquaredresid3122.055

Schwarzcriterion7.789934Loglikelihood-105.7269

Hannan-Quinncriter.7.723867F-statistic24.30524

Durbin-Watsonstat0.964078Prob(F-statistic)0.000040一、即樣本回歸方程為在顯著水平為0.05的條件下，查自由度為28-2=26的t分布，得臨界值2.06，兩參數(shù)的t值分別為1.181816,4.930035，故回歸系數(shù)均顯著不為零，回歸模型中應(yīng)包含常數(shù)項(xiàng)，X對(duì)Y有顯著影響。二、通過(guò) White檢驗(yàn)該模型中隨機(jī)誤差項(xiàng)是否存在異方差HeteroskedasticityTest:WhiteF-statistic4.502771

Prob.F(2,25)0.0214Obs*R-squared7.415120

Prob.Chi-Square(2)0.0245ScaledexplainedSS10.42735

Prob.Chi-Square(2)0.0054TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:06/10/12Time:16:39Sample:19551982Includedobservations:28VariableCoefficientStd.Errort-StatisticProb.

C8.50754993.962200.0905420.9286X-0.06378212.23389-0.0052140.9959X^20.3660030.3650621.0025780.3257R-squared0.264826

Meandependentvar111.5020AdjustedR-squared0.206012

S.D.dependentvar205.0724S.E.ofregression182.7318

Akaikeinfocriterion13.35487Sumsquaredresid834773.1

Schwarzcriterion13.49761Loglikelihood-183.9682

Hannan-Quinncriter.13.39851F-statistic4.502771

Durbin-Watsonstat1.736188Prob(F-statistic)0.021374由檢驗(yàn)結(jié)果可知，Obs*R-squared值為,7.415120明顯大于在顯著水平0.05，自由度為2的5.991，故該模型存在異方差。用加權(quán)最小二乘法剔除異方差，權(quán)重為w=(8.507549-0063782*x+0.366003*x^2)^(-0.5)所得結(jié)果為DependentVariable:YMethod:LeastSquaresDate:06/10/12Time:16:47Sample:19551982Includedobservations:28Weightingseries:WVariableCoefficientStd.Errort-StatisticProb.

C-0.1478591.484471-0.0996040.9214X1.5922710.1690469.4191350.0000WeightedStatisticsR-squared0.773361

Meandependentvar15.62696AdjustedR-squared0.764645

S.D.dependentvar7.817281S.E.ofregression6.546334

Akaikeinfocriterion6.664437Sumsquaredresid1114.217

Schwarzcriterion6.759594Loglikelihood-91.30211

Hannan-Quinncriter.6.693527F-statistic88.72011

Durbin-Watsonstat0.888837Prob(F-statistic)0.000000UnweightedStatisticsR-squared0.430297

Meandependentvar22.35857AdjustedR-squared0.408386

S.D.dependentvar14.95748S.E.ofregression11.50476

Sumsquaredresid3441.349Durbin-Watsonstat1.148584修正后再次進(jìn)行懷特檢驗(yàn)HeteroskedasticityTest:WhiteF-statistic0.653557

Prob.F(2,25)0.5289Obs*R-squared1.391228

Prob.Chi-Square(2)0.4988ScaledexplainedSS1.066437

Prob.Chi-Square(2)0.5867TestEquation:DependentVariable:WGT_RESID^2Method:LeastSquaresDate:06/10/12Time:16:50Sample:19551982Includedobservations:28CollineartestregressorsdroppedfromspecificationVariableCoefficientStd.Errort-StatisticProb.

C2461.9892805.3630.8776010.3885WGT^2-412.8241474.0004-0.8709360.3921X^2*WGT^2-17.6234720.46509-0.8611480.3973R-squared0.049687

Meandependentvar39.79345AdjustedR-squared-0.026338

S.D.dependentvar54.03521S.E.ofregression54.74218

Akaikeinfocriterion10.94410Sumsquaredresid74917.66

Schwarzcriterion11.08684Loglikelihood-150.2174

Hannan-Quinncriter.10.98774F-statistic0.653557

Durbin-Watsonstat1.624385Prob(F-statistic)0.528853結(jié)果中Obs*R-squared比原來(lái)小,故在一定程度上剔除了異方差，那么一元線(xiàn)性回歸方程為三、檢驗(yàn)誤差項(xiàng)是否存在自相關(guān)，由DW 值為0.964078，小于顯著水平為0.05,樣本容量為28下的值1.33，所以存在自相關(guān)。檢驗(yàn)是否存在高階自相關(guān)，最后結(jié)果如下：用廣義差分法消除自相關(guān)：Breusch-GodfreySerialCorrelationLMTest:F-statistic9.222670Obs*R-squared7.545722

Prob.F(1,25)0.0055

Prob.Chi-Square(1)0.0060TestEquation:DependentVariable:RESIDMethod:LeastSquaresDate:06/10/12Time:17:09Sample:19551982Includedobservations:28Presamplemissingvaluelaggedresidualssettozero.VariableCoefficientStd.Errort-StatisticProb.

C2.0850973.647450X-0.1302000.216870RESID(-1)0.5383710.1772770.5716590.5727-0.6003590.55373.0368850.0055R-squared0.269490AdjustedR-squared0.211049S.E.ofregression9.551319

Meandependentvar1.46E-15Sumsquaredresid2280.692

S.D.dependentvar10.75322Loglikelihood-101.3307

Akaikeinfocriterion7.452193F-statistic4.611335

Schwarzcriterion7.594929Prob(F-statistic)0.019739

Hannan-Quinncriter.7.495829

Durbin-Watsonstat1.757452DependentVariable:YYMethod:LeastSquaresDate:06/10/12Time:17:22Sample(adjusted):19561982Includedobservations:27afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.

C0.5118980.3673101.3936420.1757XX0.0496680.0214442.3162190.0290R-squared0.176680

Meandependentvar1.251790AdjustedR-squared0.143747

S.D