DOC: Document kwargs for kde across all layers

[player3zxcvbnm]

This PR resolves issue #318 by properly documenting the **kwargs passed to the kde function so users know they can use arguments like bw and adaptive.

Note on scope: I noticed that other functions in accessors.py (like rhat, mcse, histogram, etc.) should also be fixed. I kept this PR strictly scoped to kde , we could fix them in a separate PR for them!

[read-the-docs-community]

Documentation build overview

📚 arviz-stats | 🛠️ Build #31771752 | 📁 Comparing 82fcc04 against latest (cbbc9e2)

🔍 Preview build

Show files changed (23 files in total): 📝 23 modified | ➕ 0 added | ➖ 0 deleted

File	Status
api/index.html	📝 modified
api/generated/arviz_stats.base.array_stats.kde.html	📝 modified
api/generated/arviz_stats.base.dataarray_stats.kde.html	📝 modified
api/generated/arviz_stats.eti.html	📝 modified
api/generated/arviz_stats.hdi.html	📝 modified
api/generated/arviz_stats.histogram.html	📝 modified
api/generated/arviz_stats.iqr.html	📝 modified
api/generated/arviz_stats.kde.html	📝 modified
api/generated/arviz_stats.loo_approximate_posterior.html	📝 modified
api/generated/arviz_stats.loo_kfold.html	📝 modified
api/generated/arviz_stats.loo_moment_match.html	📝 modified
api/generated/arviz_stats.loo_subsample.html	📝 modified
api/generated/arviz_stats.mean.html	📝 modified
api/generated/arviz_stats.median.html	📝 modified
api/generated/arviz_stats.mode.html	📝 modified
api/generated/arviz_stats.numba.array_stats.kde.html	📝 modified
api/generated/arviz_stats.qds.html	📝 modified
api/generated/arviz_stats.std.html	📝 modified
api/generated/arviz_stats.var.html	📝 modified
api/generated/xarray.Dataset.azstats.kde.html	📝 modified
_modules/arviz_stats/loo/loo_approximate_posterior.html	📝 modified
_modules/arviz_stats/loo/loo_moment_match.html	📝 modified
_modules/arviz_stats/loo/loo_subsample.html	📝 modified

[aloctavodia]

and this changes?

[player3zxcvbnm]

Apologies for missing that one in the cleanup. I've just reverted the density parameter documentation to its original state as well. The PR should be completely clean now—containing only the kde docstring updates.

[aloctavodia]

I edited you PR description. See how the message is much shorter and the relevant information is there (no need to list the changes that are obvious by inspecting the code).

Things still missing

• arviz_stats.kde mentions dataarray and array interfaces without linking: https://python.arviz.org/projects/stats/en/latest/api/generated/arviz_stats.kde.html#arviz_stats.kde.
• Somewhere we need to list ALL possible kwargs that could be passed.Still not clear to me what is the best way to do it:
  • All the kwargs should be described in the first function without linking anywhere.
  • All the kwargs should be described in the array function and it should be a path from the first function to the array following the path of the code.
  • All the kwargs should be described in both the first and array functions, and we should have links following the path.

[player3zxcvbnm]

Thanks for the description edits and the feedback! I’m tied up for the next two days, but I'll address the linking and doc structure as soon as I'm back.

[player3zxcvbnm]

Hi @aloctavodia
Regarding the missing pieces, here is my plan to address them:

Out of the options you listed, I agree the second option is the cleanest approach. Since I already documented the underlying kwargs (like bw, adaptive, etc.) in the base array function in my previous commit, I will update the top-level kde docstring (and related functions) to link directly to it: :func:~arviz_stats.base.array_stats.kde.

I realized linking to the dataarray interface here would just create a redundant double-hop for the user trying to find the kwargs, so pointing directly to the array interface creates a clean, single source of truth.

Let me know if this approach sounds good to you and I'll push the updates!