search

predict-idlab / powershap

A power-full Shapley feature selection method.
Other
193 stars 18 forks source link

Release with numpy >=1.21 #21

Open deepandas11 opened 2 years ago

deepandas11 commented 2 years ago

Wondering if there's a timeline for a release that incorporates the usage of numpy>=1.21? Main motivations being:

  1. Security vulnerabilities
  2. Installing poweshap in a python environment becomes troublesome, especially when other libraries in the environment set hard requirements on numpy>1.21
jvdd commented 2 years ago

Hi @deepandas11,

Thx for your suggestion! I completely agree with your motivation. The main reason why we fixated our dependencies was to support python 3.7 (to support sklearn v0 as well).

However, as major libraries (pandas, numpy, sklearn) are dropping support for Python 3.7, I think it is fine for us to do the same (certainly in the light of security vulnerabilities).

I'll look into it in the near future :)

Cheers, Jeroen

jvdd commented 2 years ago

I updated the dependencies in PR #22 and already created a release candidate; pip install powershap==0.0.8rc1

Let me know if this works for you! :smile:

deepandas11 commented 2 years ago

I updated the dependencies in PR #22 and already created a release candidate; pip install powershap==0.0.8rc1

Let me know if this works for you! 😄

Thanks for responding so fast! Appreciate the concern.

However, this doesn't entirely fix my problem. Do let me know if the following is workable? https://github.com/predict-idlab/powershap/pull/22/files#r924985993 https://github.com/predict-idlab/powershap/pull/22/files#r925017393

jvdd commented 1 year ago

Hey @deepandas11,

Just wondering if this is still an issue? If so, can you elaborate some more on this?

Cheers, Jeroen