Florian Hartig

Hi, I just played a bit around with that - here an example of how one could modify the residuals function in piecewiseSEM to support DHARMa ... maybe one should...

I guess you would have to decide if you just want to return the scaled residuals, or a list with residuals and the DHARMa objects (as I did here). There...

I haven't worked with GEEs much, so if someone has more insights, comments would be appreciated, but based on my current understanding, I believe that this will be difficult both...

Hello Dexter, as I said above, it's not even clear to me what the right scaling would be, i.e. what residuals distributions you would expect for a "correct" GEE. If...

OK, in terms of the question: this clearly doesn't look like random, i.e. there seems to be a misfit / heteroskedasticity. What is odd though is the increasing pattern to...

I should add though that I wasn't able to re-create the problem described in #43 when simulating with a gam, possible predict.gam excludes the REs? ``` library(DHARMa) testData = createData(sampleSize...

generally true, but in the case of such a diagonal pattern, it seems hard to believe that this is a smoothing problem - one would think that it would be...

Hi all, @Istalan and @dschoenig, thanks for your comments, a few responses: 1. @Istalan, thanks for this, it's interesting to see that mgcv messes up the dispersion so profoundly, fitting...

OK, I have solved at least one problem: @Istalan, I think fit_nb$family$getTheta() just doesn't work, or is the wrong way to access the fitted theta parameter - compare the difference...

We just discussed this in some length last week in a working group at ISEC 2020. In principle, if you can simulate from your fitted model, you can create residuals...