General fixes and improvements to the documentation

microsoft / responsible-ai-toolbox-mitigations

Python library for implementing Responsible AI mitigations.

https://responsible-ai-toolbox-mitigations.readthedocs.io/en/latest/

MIT License

57 stars 6 forks source link

General fixes and improvements to the documentation #35

Closed mrfmendonca closed 2 years ago

mrfmendonca commented 2 years ago

This PR adds several fixes that were identified by Morris:

Morris made several changes to the documentation and added a lot of new comments into the notebooks
Morris fixed several grammar issues and other writing details in the docs and docstrings
Following his idea, I reformatted the case study notebooks by creating a module with common functionalities used throughout the different case study stat notebooks. This way, the case study stat notebooks now don't have a lot of repeating functions: these are all defined in a local module (this module doesn't belong to the library)
Updated the documentation (Sphinx files) for the Case Studies in order to reflect the new module used
Finally, we also fixed some errors that were occurring in a few notebooks (thanks to Morris we were able to identify them)

codecov[bot] commented 2 years ago

The author of this PR, mrfmendonca, is not an activated member of this organization on Codecov. Please activate this user on Codecov to display this PR comment. Coverage data is still being uploaded to Codecov.io for purposes of overall coverage calculations. Please don't hesitate to email us at support@codecov.io with any questions.

morrissharp commented 2 years ago

@mrfmendonca can you link the rest of the issues that will be resolved with this PR?

mrfmendonca commented 2 years ago

@mrfmendonca can you link the rest of the issues that will be resolved with this PR?

I forgot to do this. Thanks for the heads up! Now there are only a few issues left. I'll take a look at them. But they will probably be left for another PR.

morrissharp commented 2 years ago

Most everything looks great! I've gone ahead and rerun all of the notebooks with issues, but I am still receiving the same error in issue #31 in case3.ipynb. The rest of the issues appear to be resolved.

mrfmendonca commented 2 years ago

Most everything looks great! I've gone ahead and rerun all of the notebooks with issues, but I am still receiving the same error in issue #31 in case3.ipynb. The rest of the issues appear to be resolved.

I found the problem: it's because sometimes the CTGAN is loading a model that was generated in another notebook. I have already fixed it and tested it. But now I found another issue: some of the text written in the case studies mentions specific results obtained, but the results change for each run. I'll have to fix the seed to make the text in accordance with the results (this is what you mentioned in issue #29 ). I'll fix this tomorrow and push the commit to fix all these issues.