Davíd Brakenhoff

Not sure if this could be of any help, but [metran](https://github.com/pastas/metran/blob/ac3c74ceedb75447243e369b3ba771e48099f00e/metran/kalmanfilter.py#L497-L514) uses a Kalman filter and maximum likelihood to optimize the model. It is a slightly different model for decomposing...

I'm all for this change, I've run into situations where I wanted to add additional options to my response function before and this makes that possible. Also I don't think...

optional dependency statsmodels, add ps.stats.pacf()

@rubencalje , update #361 for review

@rubencalje Could you take a look at this PR :innocent: ?

I think this issue can be closed. This issue has been solved in the release of scipy 1.8. Pastas warns users when using the WellModel (which has small parameter values...

Check if implementation is correct in Pastas for small parameters

LmFit lists this on their website: https://lmfit.github.io/lmfit-py/faq.html#why-are-uncertainties-in-parameters-sometimes-not-determined

Needs more research. Check some models where this occurs to diagnose reason. Might also be reasonable not to return estimates if result is too uncertain/meaningless.

Enforce dtypes, e.g. for names (=str). Check for illegal characters in strings.