Add distribution diagnostics for BalanceDF

[neuralsorcerer]

• Added weighted EMD/CVMD/KS computation helpers and comparison functions in the weighted stats module.
• Exposed EMD/CVMD/KS BalanceDF helper methods and public comparison APIs for linked samples and direct targets.
• Added appropiate tests for EMD/CVMD/KS covering identical distributions, weighted effects, expected discrete/numeric values, validation errors, and NA-indicator skipping.

[Copilot]

Pull request overview

This PR adds three new distribution comparison metrics (Earth Mover's Distance, Cramér-von Mises distance, and Kolmogorov-Smirnov statistic) to the balance package for comparing adjusted samples to target populations. The implementation follows the established patterns for ASMD and KLD.

Changes:

• Added helper functions and three new comparison metrics (EMD, CVMD, KS) in weighted_comparisons_stats module
• Exposed EMD, CVMD, and KS methods on the BalanceDF class for both linked samples and direct target comparisons
• Added comprehensive tests covering identical distributions, weighted effects, expected values, validation errors, and NA-indicator handling

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 8 comments.

File	Description
balance/stats_and_plots/weighted_comparisons_stats.py	Implemented helper functions for weighted distributions and three new comparison metrics (emd, cvmd, ks) with validation
balance/balancedf_class.py	Added static wrapper methods and public APIs for emd, cvmd, ks on BalanceDF; removed TODO comments for these methods
tests/test_stats_and_plots.py	Added tests for emd, cvmd, ks covering identical distributions, weighted effects, expected values, validation, and NA skipping
CHANGELOG.md	Added changelog entry for the new distribution diagnostic features

Comments suppressed due to low confidence (1)

balance/stats_and_plots/weighted_comparisons_stats.py:32

• This TODO comment is now obsolete since wasserstein_distance has been imported on line 27. The comment should be removed.

# TODO: add?
# from scipy.stats import wasserstein_distance

[talgalili]

Very cool!
Please address comment of copilot, and make sure you add examples in the docstring and in the intro tutorial.
Ping me once ready for review.

[talgalili]

not: Also, in the docstring, please add links to wikipedia for each method added

[neuralsorcerer]

Very cool! Please address comment of copilot, and make sure you add examples in the docstring and in the intro tutorial. Ping me once ready for review.

@talgalili Updated :)

[Copilot]

Pull request overview

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

[talgalili]

Looking great - made some nits. Please review.

(also, please see feedback from copilot)

[Copilot]

Pull request overview

Copilot reviewed 7 out of 7 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)

balance/stats_and_plots/weighted_comparisons_stats.py:36

• This TODO comment is obsolete since wasserstein_distance is already imported at line 31. The commented import and TODO should be removed to avoid confusion.

# TODO: add?
# from scipy.stats import wasserstein_distance

[talgalili]

Pull request overview

Copilot reviewed 7 out of 7 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)
balance/stats_and_plots/weighted_comparisons_stats.py:36

• This TODO comment is obsolete since wasserstein_distance is already imported at line 31. The commented import and TODO should be removed to avoid confusion.

# TODO: add?
# from scipy.stats import wasserstein_distance

@neuralsorcerer could you please fix this nit?

[neuralsorcerer]

Pull request overview

Copilot reviewed 7 out of 7 changed files in this pull request and generated no new comments.

Comments suppressed due to low confidence (1)
balance/stats_and_plots/weighted_comparisons_stats.py:36

• This TODO comment is obsolete since wasserstein_distance is already imported at line 31. The commented import and TODO should be removed to avoid confusion.

# TODO: add?
# from scipy.stats import wasserstein_distance

@neuralsorcerer could you please fix this nit?

Updated :)

[talgalili]

Perfect, thank you very much!
I will send this to internal review.

[meta-codesync]

@talgalili has imported this pull request. If you are a Meta employee, you can view this in D90854392.

[neuralsorcerer]

@talgalili I think we’re in a good place to bump the library to 0.15.0. There have been a lot of significant improvements made.

WDYT?

[talgalili]

@talgalili I think we’re in a good place to bump the library to 0.15.0. There have been a lot of significant improvements made.

WDYT?

Fair suggestion.
I moved a bunch of issues to 0.16.0.
What I'd love to include in 0.15 is the docstring for cli, and a badge on code coverage (feel free to give me pointers, and I can also do it myself).
Once both are done, I'm happy to wrap up version 0.15.

WDYT?

[meta-codesync]

@talgalili merged this pull request in c17914a.

[neuralsorcerer]

Fair suggestion. I moved a bunch of issues to 0.16.0. What I'd love to include in 0.15 is the docstring for cli, and a badge on code coverage (feel free to give me pointers, and I can also do it myself). Once both are done, I'm happy to wrap up version 0.15.

WDYT?

I’ll work on adding CLI docstrings next, but setting up a code-coverage badge will be a bit more challenging for me as I believe we need to use an external service like coveralls or codecov (both are free to use for open-source projects) which would require extra permissions and stuff.