florianhartig
commented
1 year ago Hi skiology90,
sorry for the late reply!
Your problem is that if you have repeated observation per year, you have to group per year, i.e.
res3 = recalculateResiduals(res2, group = trainInsit$Year)
so that you have one residual per year / time point. Then, you can test for temporal autocorrelation. Grouping per location is needed if you want to test for spatial autocorrelation.
Cheers, Florian
Hi, Thanks for all of your continual efforts on DHARMa. I'm still very new to the package and having some trouble on how to evaluate only a selection of locations where repeated observations occurred.
I'm attempting to evaluate temporal correlation of the residuals from a glmer model looking at species presence/absence in streams across watersheds. The dataset (trainInsit) includes observations over the same domain from ~20 years but most we're resampled each year, meaning different stream segments (i.e. locations) were surveyed each year. However, there are approximately 1/4 of the stream segments during the 20 year period that did have repeated observations in time and those are the observations I need to consider temporal correlation.
The survey locations are defined by stream segment (COMID) and my time variable is (Year). I created a binary grouping variable (Repeated) to indicate if the COMID had any repeated observations in time.
My attempts so far have returned the below error which I assume is because I have not correctly clarified that it is only to consider Year within COMID:
Any help would be appreciated:
res = simulateResiduals(mod1) res2= recalculateResiduals(res, sel= trainInsit$Repeated == 1) res3 = recalculateResiduals(res2, group = trainInsit$COMID) testTemporalAutocorrelation(res3, time = trainInsit$Year)
Error in testTemporalAutocorrelation(res2, time = trainInsit$Year) : testing for temporal autocorrelation requires unique time values - if you have several observations per time value, either use the recalculateResiduals function to aggregate residuals per time step, or extract the residuals from the fitted object, and plot / test each of them independently for temporally repeated subgroups (typical choices would be location / subject etc.). Note that the latter must be done by hand, outside testTemporalAutocorrelation.