Conformal prediction with conditional guarantees

[Damien-Bouet]

Description

Implementation of new classes SplitCPRegressor and CCPCalibrator (and other subclasses) to implement the method proposed by Gibbs et al. (2023), and described in the issue #449

Fixes #449

Type of change

Please remove options that are irrelevant.

• New feature (non-breaking change which adds functionality)

How Has This Been Tested?

Please describe the tests that you ran to verify your changes. Provide instructions so we can reproduce. Please also list any relevant details for your test configuration

• Add tests to cover all the new features
• Still pass the old tests

Checklist

• I have read the contributing guidelines
• I have updated the HISTORY.rst and AUTHORS.rst files
• Linting passes successfully : make lint
• Typing passes successfully : make type-check
• Unit tests pass successfully : make tests
• Coverage is 100% : make coverage
• Documentation builds successfully : make doc

[thibaultcordier]

Very good first PR, thank you a lot! I give you some suggestions (format, style, code, content ...). To sum up:

• Transform PhiFunction into abstract class.
• Make the comparison with MapieRegressor to foow the same steps (check for instance).
• Remove verbose warning and duplicate checks.

[LacombeLouis]

Hey @Damien-Bouet,
Thank you and well done!
I have made some initial comments in the ccp_regression.py file. The main thing that I'm noticing is a lot of cast() and functions that seem exist in other classes. Please note these initial comments, I will have a look further regarding the other files.

[allglc]

Hello @Damien-Bouet ,
Thank you for this PR!
A lot has happened in MAPIE since this work was done (like the v1 release which changed the API). So I think quite a few modifications are necessary to make it work on the current version of the library. Do you think you have the time to look?
Otherwise we have planned to work on adaptive conformal prediction at the end of the year, so we will probably reuse some of this contribution, acknowledge your participation, and in the meantime close this PR.

[Damien-Bouet]

Hello @allglc ,
I will not currently have the time to make the necessary changes to this PR. It was however designed to be modular and compatible somehow with the MAPIE v1 which was already in discussion, so it should not be too difficult to integrate. Don't hesitate to contact me when you will start to work on adaptative CP. I will discuss about it with you with pleasure.

[allglc]

Thanks for the answer!