1、计算机1无网络环境Eviews三、实验方法与步骤(一) 数据的输入、描述及其图形处理;(二) 方程的估计;(三) 参数的检验、违背经典假定的检验;(四) 模型的处理与预测四、实验结果与数据处理实验一:中国城镇居民人均消费支出模型数据散点图:25000 20000 - 】*Y 15000 - 忙 *10000 - 八匚5000 -I 1 1 1 1 6000 8000 10000 12000 14000 16000X通过Eviews估计参数方程回归方程:Depe ndent Variable: YMethod: Least SquaresDate: 11/2 7/14 Time: 15:02Sa
2、mple: 1 31In cluded observati ons: 31VariableCoefficie ntStd. Error t-StatisticProb.1.3594770.043302 31.395250.0000C-57.90655377.7595 -0.1532890.8792R-squared0.971419Mean depe ndent var11363.69Adjusted R-squared0.970433S.D.dependent var3294.469S.E. of regressi on566.4812Akaike info15.57911criteri on
3、Sum squared resid9306127.Schwarz criteri on15.67162Log likelihood-239.4761F-statistic985.6616Durb in-Watson stat1.294974Prob(F-statistic)0.000000得出估计方程为:丫 = 1.35947661442*X - 57.9065479515异方差检验1、图示检验法15000001000000500000图形呈现离散趋势,大致判断存在异方差性。2、Park检验 LOG(E2) 16:1619.8256219.85359 0.9985910.3263LOG(X)-
4、0.9564032.204080 -0.4339240.66760.00645111.21371-0.0278092.8945952.9345685.053338249.73895.145854-76.326740.1882902.4565000.667555看到图中L0G(E2冲P值为0.6676 0.05,所以不存在异方差性3、G-Q检验ei检验: X41 1 12 124642.0282014.183 2.3046710.0439Y0.2310460.215824 1.0705300.30950.1028206796.3900.013102293.2762291.348614.33793
5、848840.214.41875-84.027581.1460340.4451460.309538e2检验:42 20 31583.4526593.4370 0.9831750.34870.6977480.040196 17.358700.96787910586.890.9646672610.864490.765515.380822408507.15.46164-90.28493301.32452.748144第一个图中的残差平方和为 848840.2第二个图中的残差平方和为 2408507,所以不存在异方差性所以 F 值为 2408507/848840.2 = 2.8374 通过四种不同的检
6、验得知除了图示检验法得出异方差的结论, 其他的检验的结论都是不存在异方差的。5、WLS(加权最小二乘法)修正 17:14Weight ing series: E3-85.6942624.15675 -3.5474250.00131.3622210.002307 590.5615Weighted Statistics1.00000013474.5361353.7427.932649.55981022626.739.652325-146.1770348762.92.061818Un weightedStatistics0.9714130.970427566.54159308110.2.178992
7、实验二:中国粮食生产函数1、回归方程 LOG(Y) 12/11/14 Time:06 1983 2007 25LOG(X1)0.3811450.0502427.586182LOG(X2)1.2222890.1351799.042030LOG(X3)-0.0811100.015304-5.300024LOG(X4)-0.0472290.044767-1.0549800.3047LOG(X5)-0.1011740.057687-1.7538530.0956-4.1731741.923624-2.1694340.04290.98159710.709050.9767530.0933960.014240
8、-5.4599680.003853-5.16743874.24960202.68261.791427得出回归方程为:LOG(Y) = 0.381144581612*LOG(X1) + 1.22228859801*LOG(X2) - 0.0811098881534*L0G(X3)- 0.04722870996*L0G(X4) - 0.101173736285*LOG(X5) - 4.173*通过检验结果可知 RR较大且接近于1,而且F=202.6826 Fo.o5(5,19) = 2.74 ,故认为粮食产量与上述变量之间总体线性关系显著。但是由于其中 X4、X5前的参数估计值未通过t检验,且符号
9、的经济意义不合理,故认为解释变量之间存在多重 共线。2、相关系数表LNX1LNX2LNX3LNX4LNX5-0.5687440.4517000.9643570.440205-0.214097-0.697625-0.0732700.3987800.4112790.279528由表可知LnX与LnX2之间存在高度的线性相关性3、简单的回归形式LnY 与 LnX LNY150.2240050.025515 8.7792938.9020080.206034 43.206570.770175Mea n depe ndent var0.7601820.045737-3.2551890.048114-3.1
10、5767942.6898677.075990.939435LnY 与 LnX2-0.3834340.509669 -0.7523210.459515.157485.912971 2.5634290.01740.024017-0.0184170.094252-1.8090630.204321-1.71155324.613290.5659860.3352190.459489LnY 与 LnX3180.1080670.0852711.2673350.21779.6197220.85974411.189050.0652740.0246340.092239-1.8522550.195684-1.7547
11、4525.153191.6061390.5977490.217717LnY 与 LnX40.1669760.028274 5.9056708.9490900.298255 30.004790.6026050.5853270.060143-2.7075780.083194-2.61006835.8447234.876930.6255280.000005LnY 与 Ln%Depe nde nt Variable:190.4887310.234606 2.0831990.04855.6007492.452207 2.2839620.03190.1587330.1221560.176118 Schwarz criterion26.46999 F-statistic0.327932 Prob(F-statistic)比较各个回归方程的R2可知Y与X1的氏最大,即粮食生产受农业化肥施用量最大, 与经验相符,因此选为初始的回归方程。且初始化回归方程为:LOG(Y) = 0.224004867873*LOG(X1) + 8.902008217844、逐步回归
copyright@ 2008-2023 冰点文库 网站版权所有
经营许可证编号:鄂ICP备19020893号-2