[SciPy-User] confidence interval for leastsq fit
Till Stensitzki
mail.till at gmx.de
Thu Apr 28 12:23:41 EDT 2011
> If you can estimate this fixed parameter, then you really need to
> include in your model. That type of 'messing about' does not provide the
> correct values so of course your confidence intervals will be 'wrong'.
> Wrong because the variances and covariances are *conditional* on the
> fixed value and do not account for the uncertainty in estimating the
> 'fixed' parameter.
Please read the reference, its only half a page long, maybe my explanation
weren't clear enough. The principal of method is, that as long the chi^2
values of a parameter set are not significantly different from the
chi^2 value of the optimal set (in a statistical sense, can be
estimated by using the fisher-distribution.) the parameters are "valid".
More information about the SciPy-User
mailing list