DOC/DEPR: port pandas.rpy to rpy2 guide

[jorisvandenbossche]

Addresses the doc part of #9602

[TomAugspurger]

Thanks for this. Thoughts on adding a section explaining why we're deprecating it? (no real maintainers, rpy2 is better... I think there were some others).

[jorisvandenbossche]

Yes, indeed, good point. I was also thinking that, but in the end just wrote that it is "similar" functionality.

I think the main reason is to "avoid duplication"? (having the same functionality in both pandas and rpy2 is not needed, and anyhow, those who used pandas.rpy were in fact already using rp2)
Saying that rpy2 is better is not yet fully correct .. as there are still some things better in the pandas version to be ported (eg NaN handling, see the issue).

[TomAugspurger]

Similar is probably good enough. I just didn't read closely enough. 👍

[ARF1]

@jorisvandenbossche I just stumbled across this PR in the Travis CI build queue and was wondering how one would convert a pandas dataframe to a R dataframe with rpy2. Any chance the docs could contain a minimal how-to for this very frequent use-case?

[jorisvandenbossche]

The pandas2ri.py2ri function as mentioned in the new docs in this PR (or even only doing pandas2ri.activate() will be enough). But I assumed knowledge of the existing functions, so this is equivalent to the convert_to_r_dataframe.
But it would be nice to also have docs geared towards new users (who didn't know the old way). But let's say that is for another PR.

[jreback]

lgtm. merge when ready.