Plots of residuals from simulated time series data sets generated with differing levels of correlation $\rho$ between error terms for adjacent time points. |
另一个要求是:the error terms have a constant variance, $Var(\epsilon_i)=\sigma^2$. 对于non-constant variances in the errors, or heteroscedasticity,如果在residual plot图中出现了funnel shape的形状时,那说明error variance的一致性出现问题。对于这种情形,我们可以通过对因变量Y施加凹形函数,将突出的residual plot图形往回拉平。这样的函数有$logY, \sqrt{Y}$等。
No comments:
Post a Comment