iancovert / sage

For calculating global feature importance using Shapley values.
MIT License
255 stars 34 forks source link

Implement parallel PermutationEstimator #15

Closed j-adamczyk closed 1 year ago

j-adamczyk commented 2 years ago

Solves Issue #11, adding parallelization to PermutationEstimator with Joblib. It is controlled via standard n_jobs argument, with previous sequential behavior as default.

Additionally, determinism is ensured via standard random_state argument, and it has been added to all estimators. Results are highly consistent, which can be checked in consistency.ipynb notebook.