命令:
reg y x1 x2 x3
predict da,residual
求出残差后我检验了一下,发现用回归系数乘以x1 x2 x3得到的y加上残差后和进行reg的y并不相等。请问这是怎么回事啊。
Source | SS df MS Number of obs = 725
-------------+------------------------------ F( 3, 721) = 98.59
Model | 34.7526737 3 11.5842246 Prob > F = 0.0000
Residual | 84.7207719 721 .117504538 R-squared = 0.2909
-------------+------------------------------ Adj R-squared = 0.2879
Total | 119.473446 724 .165018571 Root MSE = .34279
------------------------------------------------------------------------------
y | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
x1 | -4.36e+07 7585710 -5.74 0.000 -5.85e+07 -2.87e+07
x2 | .3251838 .0201112 16.17 0.000 .2857004 .3646673
x3 | -.1711691 .0379364 -4.51 0.000 -.2456482 -.09669
_cons | -.0446073 .0188537 -2.37 0.018 -.081622 -.0075925
------------------------------------------------------------------------------
predict e,residual
我验证了一组数据e
y | x1 | x2 | x3 |
-0.99503 | 4.22553E-11 | 0.165286 | 0.413776 |
e:-.2518547
-4.36e+07 *4.22553E-11+.3251838 *0.165286+-.1711691 *0.413776-.2518547
结果不等于-0.99503