Coefficient Std. Error t-Statistic Prob.
C(1) 4.604784 0.092428 49.82044 0.0000
C(2) -5.46E-05 5.99E-05 -0.912343 0.3622
C(3) 3.41E-08 1.24E-08 2.744363 0.0063
C(4) -2.67E-12 8.22E-13 -3.250273 0.0013
R-squared 0.727830 Mean dependent var 4.776104
Adjusted R-squared 0.725726 S.D. dependent var 0.096750
S.E. of regression 0.050669 Akaike info criterion -3.116850
Sum squared resid 0.996135 Schwarz criterion -3.076326
Log likelihood 614.9025 F-statistic 345.8602
Durbin-Watson stat 1.096628 Prob(F-statistic) 0.000000
随着样本个数的增加R-squared和Adjust R增加,但是t-Statistic增加也很快,请问各位大牛这对我的回归方程是否有重要影响.
还有就是我在书上看到拟合度很好的方程式但Log likelihood确是负数. 一般理论不是残差越小,L值越大,越大说明模型越正确吗?