COVARIANCE¶
An optional section that requests PoPy to compute Standard Errors of the f[X] variables, after running a Fit Script.
The COVARIANCE option is only available in a Fit Script or Tut Script. The Tut Script simply passes the COVARIANCE section to the Fit Script. The MFit Script and MTut Script do not currently include this option.
COVARIANCE Example¶
If you miss out the ‘COVARIANCE’ field, that is equivalent to putting:-
COVARIANCE: {NO_COVARIANCE: {}}
In the Fit Script and no standard errors are computed.
If you use this setting:-
COVARIANCE: {SANDWICH: {}}
After the Fit Script outputs the final f[X] estimates, PoPy will compute the standard errors utilising a sandwich operator. See Standard Errors for more details.
By default PoPy computes standard errors using only the main f[X] estimates and excludes the variance f[X].
You can include the variance f[X] using the following setting:-
COVARIANCE: {SANDWICH: {var_covariance_flag: True}}
This will compute standard error estimates for all f[X] however it will take a long time if you have large variance f[X] matrices in your model. Also note that the standard error estimates will be different if the variances are included in the computation.