DOC: groupby by indexer to 'resample' data

[tomchor]

Related to this question.

I would like to propose either to expand .resample to work on indexes that are integers or to create another method that does the same thing (maybe more general .upsample and .downsample methods).

If it was implemented as an expansion of resample I imagine something like:

df = pd.DataFrame(np.random.randn(10,2))
df.resample(5, np.std)

Which would produce:

          0         1
0  1.184582  0.492113
5  0.533134  0.982562

which is a pretty forward downsample case. For upsampling maybe some discussion is needed on how to return the indexes.

This should be simple to program and I think it would be a very useful shortcut.

[jreback]

You mean like this?

In [6]: df.groupby(df.index / 5).std()
Out[6]: 
          0         1
0  0.324344  0.824444
1  1.088061  0.854456

[jreback]

So'll rerpupose this to be a doc issue, if you'd like to add an example to: http://pandas.pydata.org/pandas-docs/stable/groupby.html#examples

[tomchor]

Yes, that is exactly what I meant! Nobody in SO that saw the question (including me) could come up with that.

I'm new to github, but I'll look up how to contribute and I'll probably write an example of that.

Thanks a lot.

[jreback]

contributing docs are here

thanks!

[pfrcks]

@jreback Hi, I was working on this and have implemented the requirement. However while doing so
I utilized
python make.py clean
python make.py --no-api

Followed by

python make.py --single groupby
As the example was added to groupby.
The contribution guide says "you must make sure not to commit the file deletions to your Git repository!"

Currently upon executing git status, it shows all of the rst files it deleted. I am unsure what to do now. Should I simply remove those deleted rst files being tracked or something else?

Thanks.

[jreback]

just commit that single file
the git stash to remove the other changes