《第四章计量经济学实验报告.doc》由会员分享，可在线阅读，更多相关《第四章计量经济学实验报告.doc（12页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、计量经济学实验报告二一二年三月实验名称多元线性回归模型实验日期2012-04-28实验目的 为了规划中国未来旅游产业的发展，需要定量定分析影响中国旅游市场发展的主要因素实验内容1、 建立工作文件并导入数据2、 模型设定3、 估计参数4、 修正多重共线性实验过程及其处理（数据、图表、方案等）：X3的方差扩大因子是：Scalar VIFJZZ=15.1.导入数据建立工作文件 首先，双击EViews图标，进入EViews主页。在菜单中依次点击FileNweWorkfile,出现对话框“Workfile Range”.选择“Annual”,并在“Start date”中输入开始时间或序号，点击“OK”
2、 导入数据 打开“Eviews”主页，点击“File”“import” “Read Text Lotus Excel”在workfile中，选中x和y，右击鼠标，选中“open”“as grup”即可。 obsYX2X3X4X5X619941023.50052400.00414.700054.90000111.78005.19951375.70062900.00464.000061.50000115.70005.19961638.40063900.00534.100070.50000118.58006.19972112.70064400.00599.8000145.7000122.64006.
3、19982391.20069450.00607.0000197.0000127.85006.19992831.90071900.00614.8000249.5000135.17006.20003175.50074400.00678.6000226.6000140.27006.20013522.40078400.00708.3000212.7000169.80007.20023878.40087800.00739.7000209.1000176.52007.20033442.30087000.00684.9000200.0000180.98007.20044710.700.0731.800021
4、0.2000187.07007.20055285.900.0737.1000227.6000193.05007.20066229.740.0766.4000221.9000345.70007.20077770.620.0906.9000222.5000358.37007.2.模型设定Yt=1+2X2+3X3+4X4+5X5+6X6+3.参数估计对Yt X2 X3 X4 X5 X6的数据进行OLS回归Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:34Sample: 1994 2007Included obser
5、vations: 14VariableCoefficientStd. Errort-StatisticProb.C-1471.9561137.046-1.0.2316X20.0.9.0.0000X34.1.4.0.0031X42.1.2.0.0283X51.1.1.0.3436X6-354.9821244.8486-1.0.1852R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression127.4135Akaike info criterion12.
6、83028Sum squared resid.5Schwarz criterion13.10416Log likelihood-83.81195F-statistic593.4168Durbin-Watson stat1.Prob(F-statistic)0. 由此可见，该模型R2=0.9956可决系数很高，F检验值593.4168，明显显著。但是当a=0.05时，ta/2(n-k)=t0.025(14-6)=2.31,不仅X5、X6的系数的符号与预期相反，这表明很可能存在严重的多重共线性。 计算各解释变量的相关系数，选择X2、X3、X4、X5、X6数据，点“view/correlation”
7、的相关系数矩阵。 表： 相关系数矩阵变量X2X3X4X5X6X21.0.0.0.0.X30.1.0.0.0.X40.0.1.0.0.X50.0.0.1.0.X60.0.0.0.1.修正多重共线性 采用逐步回归的办法，去检验和解决多重共线性问题。分别作Y对X2、X3、X3、X4、X5、X6的一元回归Y=C +2X2+Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-1700.756303.4618-5.0.0001X20.0.18.248830.0000R-squared0.Mean dependent
8、 var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression374.1475Akaike info criterion14.81874Sum squared resid.Schwarz criterion14.91003Log likelihood-101.7312F-statistic333.0199Durbin-Watson stat0.Prob(F-statistic)0.Y=C+3X3+Dependent Variable: YMethod: Least SquaresDate: 01/01/0
9、4 Time: 01:01Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-5675.0481006.125-5.0.0001X314.022451.9.0.0000R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression699.6807Akaike info criterion16.07069Sum squared res
10、id.Schwarz criterion16.16198Log likelihood-110.4948F-statistic86.65749Durbin-Watson stat0.Prob(F-statistic)0.Y=C+4X4+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:03Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C12.355911143.2480.0.9916X41
11、9.610305.3.0.0067R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression1458.671Akaike info criterion17.54000Sum squared residSchwarz criterion17.63130Log likelihood-120.7800F-statistic10.69943Durbin-Watson stat0.Prob(F-statistic)0.Y=C+5X5+Dependent Vari
12、able: YMethod: Least SquaresDate: 01/01/04 Time: 01:06Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-480.5047501.1310-0.0.3566X522.595722.8.0.0000R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression741.5289Aka
13、ike info criterion16.18687Sum squared resid.Schwarz criterion16.27816Log likelihood-111.3081F-statistic75.83622Durbin-Watson stat1.Prob(F-statistic)0.Y=C+6X6+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:07Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-S
14、tatisticProb.C-17474.792305.914-7.0.0000X63025.062330.99779.0.0000R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression711.0586Akaike info criterion16.10295Sum squared resid.Schwarz criterion16.19424Log likelihood-110.7207F-statistic83.52542Durbin-Wats
15、on stat0.Prob(F-statistic)0.一元回归估计结果变量 X2 X3 X4 X5 X6参数估计值 0.0588 14.0225 19.6103 22.5957 3025.062T统计量 18.2488 9.3090 3.2710 8.7084 9.1392R2 0.9652 0.8784 0.4714 0.8634 0.8744调整R2 0.9623 0.8682 0.4273 0.8520 0.8639其中，加入X2的方程调整R2最大，以X2为基础，顺次加入其它变量逐步回归Y=C+2X2+3X3+Dependent Variable: YMethod: Least Squ
16、aresDate: 01/01/04 Time: 01:10Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-3489.548265.1268-13.161810.0000X20.0.15.263450.0000X35.0.7.0.0000R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression155.1743Akaike
17、info criterion13.11438Sum squared resid.7Schwarz criterion13.25133Log likelihood-88.80069F-statistic997.4059Durbin-Watson stat1.Prob(F-statistic)0.Y=C+2X2+4X4+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:13Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-
18、StatisticProb.C-2105.420184.1656-11.432210.0000X20.0.24.210410.0000X45.1.5.0.0002R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression206.7799Akaike info criterion13.68860Sum squared resid.4Schwarz criterion13.82554Log likelihood-92.82018F-statistic559
19、.2848Durbin-Watson stat1.Prob(F-statistic)0.Y=C+2X2+5X5+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:15Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-1699.171340.3756-4.0.0004X20.0.5.0.0001X50.4.0.0.9900R-squared0.Mean dependent var3527.
20、783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression390.7814Akaike info criterion14.96158Sum squared resid.Schwarz criterion15.09852Log likelihood-101.7311F-statistic152.6364Durbin-Watson stat0.Prob(F-statistic)0.Y=C+2X2+6X6+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Tim
21、e: 01:17Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-6821.4201579.554-4.0.0012X20.0.8.0.0000X6935.0066285.46353.0.0074R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression278.0490Akaike info criterion14.28088
22、Sum squared resid.6Schwarz criterion14.41782Log likelihood-96.96617F-statistic306.8613Durbin-Watson stat1.Prob(F-statistic)0. 二元回归加入新变量的回归结果变量X2X3X4X5X60调整R2X2，X30.0410(15.2635)5.1427(7.6657)0.9935X2，X40.0523(24.21041)5.4830(5.3186)0.9885X2, X50.0587(5.6753)0.0536(0.0128)0.9589X2,X60.0434(8.2145)935
23、.0066(3.2754)0.9792经比较，新加入点X方程调整R2=0.9935，改进最大，而且各参数的t检验显著，选择保留X3，再加入其它新的保留逐步回归。Y=C+2X2+3X3+4X4+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:19Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-3136.713295.9214-10.599820.0000X20.0.16.041780.
24、0000X33.0.3.0.0033X42.1.1.0.0766R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression138.0499Akaike info criterion12.92806Sum squared resid.8Schwarz criterion13.11065Log likelihood-86.49645F-statistic841.4324Durbin-Watson stat1.Prob(F-statistic)0.Y=C+2
25、X2+3X3+5X5+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:22Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-3468.894272.6299-12.723820.0000X20.0.7.0.0000X35.0.7.0.0000X51.1.0.0.4877R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.
26、D. dependent var1927.495S.E. of regression158.6812Akaike info criterion13.20663Sum squared resid.4Schwarz criterion13.38922Log likelihood-88.44640F-statistic636.0442Durbin-Watson stat1.Prob(F-statistic)0.Y=C+2X2+3X3+6X6+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:24Sample: 1994
27、 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-2723.9511240.830-2.0.0529X20.0.13.702110.0000X35.1.4.0.0007X6-178.7471282.6205-0.0.5413R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression159.5877Akaike info criterion13.218
28、02Sum squared resid.2Schwarz criterion13.40061Log likelihood-88.52614F-statistic628.8019Durbin-Watson stat1.Prob(F-statistic)0.三元回归结果变量X2 X3X4X5X6调整R2X2，X3，X40.0435(16.0418)3.6660(3.8314)2.1786(1.9744)0.9949X2，X3，X50.0379(7.5541)5.1881(7.5308)1.2342(0.7205)0.9932X2，X3，X60.0418(13.7021)5.7560(4.8365)
29、-178.7471(-0.6325)0.9931 在X2、X3基础上加入X4后的方程调整R2有所改善，且各参数的t检验都显著。而加入X5时，调整R2有所下降，且X5参数的t检验变得不显著。加入X6后，调整R2有所下降，X6参数的符号也变得不合理。保留X4，在加入其它新变量逐步回归。Y=C+2X2+3X3+4X4+5X5+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:25Sample: 1994 2007Included observations: 14VariableCoefficientStd. Error
30、t-StatisticProb.C-3070.421294.5183-10.425230.0000X20.0.9.0.0000X33.0.3.0.0041X42.1.2.0.0558X51.1.1.0.2579R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression134.9883Akaike info criterion12.92071Sum squared resid.6Schwarz criterion13.14894Log likelihoo
31、d-85.44494F-statistic660.3898Durbin-Watson stat1.Prob(F-statistic)0.Y=C+2x2+3x3+4x4+6x6+Dependent Variable: YMethod: Least SquaresDate: 01/01/04 Time: 01:28Sample: 1994 2007Included observations: 14VariableCoefficientStd. Errort-StatisticProb.C-1329.3491129.003-1.0.2692X20.0.15.629540.0000X34.1.4.0.
32、0018X42.1.2.0.0296X6-398.0537241.2550-1.0.1334R-squared0.Mean dependent var3527.783Adjusted R-squared0.S.D. dependent var1927.495S.E. of regression127.5059Akaike info criterion12.80666Sum squared resid.8Schwarz criterion13.03489Log likelihood-84.64659F-statistic740.4428Durbin-Watson stat1.Prob(F-sta
33、tistic)0.四元回归结果加入新变量的回归结果变量X2X3X4X5X6调整R2X2，X3，X4,X50.0394（9.1108）3.5794（3.8145）2.4034（2.1951）1.7859（1.2078）0.9951X2，X3，X4，X60.0461（15.6295）4.6031（4.3817）2.8112（2.5817）-398.0537（-1.6499）0.9956 当加入X5时，调整R2有所增加，但其参数的t检验不显著。加入X6后，调整R2也有所增加，的其参数的t检验不显著，且参数为负值不合理。从相关销售也可看出，X5、X6引起了多重共线性，予以剔除。在表中，虽然X4参数的t
34、值为1.9744，其P值为0.0766，在a=0.05显著性水平下线性不为0，可予以保留。 实验结果及分析：最后修正严重多重共线性影响后的回归结果为： Yt=-3136.713+0.0435X3t+3.6660X3t+2.1786X4t T=(-10.5998) (16.0418)(3.8314) (1.9744)R2=0.9661 调整R2=0.9949 F=841.4324 DW=1.1763 这说明，在其它因素不变的情况下，当国内旅游人数X2每增加1万人次，城镇居民人均旅游花费X3和农村居民人均旅游花费X4分别增加1元时，平均说来国内旅游收入Yt将增加0.0435亿元、3.666亿元和2.1786亿元。指导教师简评：成绩：