Appearing in predict_gfbootstrap.R, sometimes all importances are 0,
for all predictors. This seems broken, it shouldn't happen.
Why are the importances 0.
For now, setting combined objects that behave that way to NA. Leaving
issue open, because it may be a bug, possibly in gradientForest.
The object I noticed it in was a demersal trophic level, epi
depth, and all surveys. However, demersal trophic level is only
found in watson survey, so this is the equivalent to just the
demersalepiwatson survey. demersal and epi suggests that
the survey may be empty.
Appearing in
predict_gfbootstrap.R, sometimes all importances are 0, for all predictors. This seems broken, it shouldn't happen.Why are the importances 0.
For now, setting combined objects that behave that way to NA. Leaving issue open, because it may be a bug, possibly in gradientForest.
The object I noticed it in was a
demersaltrophic level,epidepth, andallsurveys. However,demersaltrophic level is only found inwatsonsurvey, so this is the equivalent to just thedemersalepiwatsonsurvey.demersalandepisuggests that the survey may be empty.