第六章作业Micrdiosoft PowerPoint 演示文稿.ppt

第六章作业 3 已知某行业的年销售额（X，万元），以及该行业内某公司的年销售额（Y，万元），数据如下： （1）建立回归模 （2）观察残差图 （3）计算DW统计量 （4）用差分法和广义差分法消除自相关 Dependent Variable: Y Method: Least Squares Date: 05/06/10 Time: 17:46 Sample: 1975 1994 Included observations: 20 Variable Coefficient Std. Error t-Statistic Prob. C -1.454750 0.214146 -6.793261 0.0000 X 0.176283 0.001445 122.0170 0.0000 R-squared 0.998792 Mean dependent var 24.56900 Adjusted R-squared 0.998725S. D. dependent var 2.410396 S.E. of regression 0.086056 Akaike info criterion -1.972991 Sum squared resid 0.133302 Schwarz criterion -1.873418 Log likelihood 21.72991 F-statistic 14888.14 Durbin-Watson stat 0.734726 Prob(F-statistic) 0.000000 回归模型 观察残差图 DW统计量 LM检验 Breusch-Godfrey Serial Correlation LM Test: F-statistic 5.864187 Probability 0.012290 Obs*R-squared 8.459475 Probability 0.014556 Test Equation: Dependent Variable: RESID Method: Least Squares Date: 05/06/10 Time: 17:50 Variable Coefficient Std. Error t-Statistic Prob. C 0.034823 0.176548 0.197246 0.8461 X -0.000239 0.001192 -0.200550 0.8436 RESID(-1) 0.759891 0.244378 3.109491 0.0067 RESID(-2) -0.200584 0.250830 -0.799680 0.4356 R-squared 0.422974 Mean dependent var -8.50E-16 Adjusted R-squared 0.314781 S.D. dependent var 0.083761 S.E. of regression 0.069336 Akaike info criterion -2.322859 Sum squared resid 0.076919 Schwarz criterion -2.123712 Log likelihood 27.22859 F-statistic 3.909458 Durbin-Watson stat 2.124566 Prob(F-statistic) 0.028590 结果 差分法 差分：变量的本期值与以前某期值之差。一阶差分法：变量的本期值与前一期值之差。一阶差分法简单易行，在实践上有较广泛的应用。 一般只要模型的DW值小于拟合优度值，就可以使用一阶差分法。 一阶差分法回归结果 Dependent Variable: Y1 Method: Least Squares Date: 05/06/10 Time: 18:33 Sample(adjusted): 1976 1994 Included observations: 19 after adjusting endpoints Variable Coefficient Std. Error t-Statistic Prob. C