Open ngupta20 opened 1 year ago
Hi Nandish, thanks for raising this issue and providing some temporary resolutions. I just checked shap
GitHub page, and found out some recent issues related to additivity, e.g., https://github.com/slundberg/shap/issues/2778 and https://github.com/slundberg/shap/issues/2777. Just curious whether you have tried your dataset on shap
as well and whether the additivity issue still exist. Thanks!
Description
Following the
0.1.3
release, the additivity of my logit-linked xgboost local tree shaps broke. I found that pinning mynumpy
version back to either1.21.6
or1.22.4
temporarily solved the issue. I also tested everynumpy
version from1.23.1
to1.23.5
and found that allnumpy
versions1.23.*
were associated with the value changes.Brainstorming
It makes me wonder if there is any
numpy
ornumpy
-adjacent usage in FastTreeSHAP that could be reimplemented to be compatible with new versions. I would be glad to contribute if that's the case and someone could help restrict the scope of the problem (if I can't restrict it myself quickly enough). As we all know from #14, that would be highly preferable to pinning numpy versions. My instinct tells me this issue may link to an existing NumPy update note or issue... haven't looked too deeply yet.Good Release
In any case, I still think the decision to relax the
numpy
constraint is the right one and anyone who runs into the same behavior can just pin the version themselves in the short term.Environment Info
Partial result of
pip list
uname --all
lscpu
Thanks for your responsiveness to the community!
Nandish Gupta Senior AI Engineer, SolasAI