Ian Covert

Hi Domenico, thanks for posting and I'm glad you enjoyed the paper and blog post. You're right that we're not currently doing any parallelization for SAGE value estimation - we...

Hi Jakub - thanks for following up on this, and sorry for the slow reply. I've thought about this a bit (though I haven't had a chance to work on...

I think you're right, it should be similar to how you would parallelize SHAP. Very cool, well please let me know if I can be of any help with your...

Hi there, sorry for the slow reply! The easiest way right now would be to save the objects that are necessary to construct the estimator (in many cases, just by...

Hi Jakub - sorry for the slow reply. I think allowing `relaxed=True` is a nice feature, but it's probably not necessary. The feature isn't described in any of the demos...

Setting `n_jobs=1` as a default argument sounds like the right approach. And making `joblib` a required dependency seems like a good idea to me - like you said, it should...

Okay, so it would be new for you as well. Well I appreciate your willingness to look into this, I'll be happy to help in any way I can.

Closing this because Jakub's PR has been merged. Thanks!

That's a good question, we never tested FastSHAP for regression models but I can explain how I would do it. Recall that the two steps for our approach are 1)...