实验参考结果.docVIP

实验3参考结果 数据集Table2-8： 回归结果如下： Dependent Variable: FOODEXP Method: Least Squares Sample: 1 55 Included observations: 55 Variable Coefficient Std. Error t-Statistic Prob.?? C 94.20878 50.85635 1.852449 0.0695 TOTALEXP 0.436809 0.078323 5.577047 0.0000 R-squared 0.369824 ????Mean dependent var 373.3455 Adjusted R-squared 0.357934 ????S.D. dependent var 83.43510 S.E. of regression 66.85575 ????Akaike info criterion 11.27864 Sum squared resid 236893.6 ????Schwarz criterion 11.35163 Log likelihood -308.1625 ????F-statistic 31.10345 Durbin-Watson stat 2.083299 ????Prob(F-statistic) 0.000001 回归中得到的残差看起来如下图所示： 将残差对总支出描点，我们看到： 看来，随着总支出的增加，残差的绝对值也在增加，但可能是非线性的。 帕克检验、格里瑟检验、GQ检验和怀特检验都表明异方差性的存在。具体过程略。 以总支出的倒数为权重，采用加权最小二乘法，可以剔除异方差性。结果如下： Dependent Variable: FOODEXP Method: Least Squares Sample: 1 55 Included observations: 55 Weighting series: W＝1／totalexp Variable Coefficient Std. Error t-Statistic Prob.?? C 76.54388 37.94352 2.017311 0.0487 TOTALEXP 0.464997 0.063171 7.360907 0.0000 Weighted Statistics R-squared 0.071308 ????Mean dependent var 362.3459 Adjusted R-squared 0.053786 ????S.D. dependent var 62.78915 S.E. of regression 61.07723 ????Akaike info criterion 11.09784 Sum squared resid 197712.7 ????Schwarz criterion 11.17084 Log likelihood -303.1906 ????F-statistic 54.18295 Durbin-Watson stat 2.122781 ????Prob(F-statistic) 0.000000 Unweighted Statistics R-squared 0.368266 ????Mean dependent var 373.3455 Adjusted R-squared 0.356346 ????S.D. dependent var 83.43510 S.E. of regression 66.93833 ????Sum squared resid 237479.3 Durbin-Watson stat 2.078885 （5）采用对数模型，结果如下。使用怀特检验表明没有异方差性。这个例子表明，对数变换通常能够减少异方差性。因此，判断是否存在异方差性时，回归模型所采用的函数形式也至关重要。 Dependent Variable: LOG(FOODEXP) Method: Least Squares Sample: 1 55 Included observations: 55 Variable Coeffi