search

aws / amazon-sagemaker-clarify

Fairness Aware Machine Learning. Bias detection and mitigation for datasets and models.
Apache License 2.0
66 stars 38 forks source link

chore: add support for python 3.10 in smclarify #139

Closed Satish615 closed 1 year ago

Satish615 commented 1 year ago

Description of changes:

  1. update smclairfy to 0.4 to support api changes the bias_stat metrics and model performance report.
  2. update github workflows to build the packages on python 3.7, 3.8, 3.9, 3.10and run the test-coverage matrix for python 3.9 and 3.10
  3. update the read.me, setup.py , setup.cfg to support python 3.10.
  4. fix ValueError with sns.countplot() in example notebooks. 
    ValueError: The truth value of a Series is ambiguous. Use a.empty, a.bool(), a.item(), a.any() or a.all().

Testing: ./devtool all passed in local. github actions workflow passed for 3.7, 3.8 and 3.9, 3.10 on the forked repo.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

goswamig commented 1 year ago

BTW I noticed we don't have coverage for all bias metrics in our test here we have total 21 bias metrics (pre and post train).

but report reflect 19 or 20 https://github.com/Satish615/amazon-sagemaker-clarify/actions/runs/4410316162/jobs/7727661324 ?

Satish615 commented 1 year ago

BTW I noticed we don't have coverage for all bias metrics in our test here we have total 21 bias metrics (pre and post train).

but report reflect 19 or 20 https://github.com/Satish615/amazon-sagemaker-clarify/actions/runs/4410316162/jobs/7727661324 ?

Nice catch! I have verfied that we don't have unit test coverage for 3 metrics TVD(pre), CDDPL(post), DPPL (post). I am adding a tracking sim for these.