Koalas: Inline method docs, rather than using "derived_from"

[rxin]

The problem is that the equivalent methods in pandas usually have a lot of arguments, and some of them don't make to support in Koalas or are temporarily not supported. Showing those in the examples are simply confusing to end users when they read the docs. We should just inline the docs. We can however create them using pandas docs as the baseline.

Here is an initial list:

frame.py-    def iteritems(self):
frame.py-    def to_html(self, buf=None, columns=None, col_space=None, header=True, index=True,
frame.py-    def isnull(self):
frame.py-    def notnull(self):
frame.py-    def toPandas(self):
frame.py-    def assign(self, **kwargs):
frame.py-    def dropna(self, axis=0, how='any', thresh=None, subset=None, inplace=False):
frame.py-    def drop(self, labels, axis=0, errors='raise'):
frame.py-    def get(self, key, default=None):
frame.py-    def pipe(self, func, *args, **kwargs):

series.py-    def rename(self, index=None, **kwargs):
series.py-    def reset_index(self, level=None, drop=False, name=None, inplace=False):
series.py-    def isnull(self):
series.py-    def notnull(self):
series.py-    def dropna(self, axis=0, inplace=False, **kwargs):
series.py-    def head(self, n=5):
series.py-    def value_counts(self, normalize=False, sort=True, ascending=False, bins=None, dropna=True):

[rxin]

cc @thunterdb what do you think about this one?

[thunterdb]

I totally agree, we should not need to use derived_from. Personnally, I would rather have (slightly misleading) doc from pandas for now rather than no doc at all.

This is a great starter task if someone wants to do it on their favorite functions.

[rxin]

A lot of them are probably blocked by https://github.com/databricks/koalas/issues/168

[abishekganesh72]

Hi let me know if I can be of any help to for this

[thunterdb]

@abishekganesh72 you are welcome to help. Here is an example, taking Series.value_counts:
It would be great to take the documentation from pandas:
https://github.com/pandas-dev/pandas/blob/v0.24.2/pandas/core/base.py#L1222
and copy the docstring into series.py:
https://github.com/databricks/koalas/blob/master/databricks/koalas/series.py#L321

Some changes will have to be made:

• removing the argument that are not available yet (like bins) and making note in the doc that it is not supported yet
• removing the example that uses bins
• making sure that the tests in the docstring run

I suggest starting with one function to build confidence in the process. Later we can batch multiple functions at once.

[abishekganesh72]

Sounds good, let me take a crack at it.

[abishekganesh72]

@thunterdb I am trying to push my code the repo but I am not authorized to do so I guess. Kindly let me know how am I supposed to push my changes to repo.

[rxin]

You will need to fork the repository first
https://help.github.com/articles/fork-a-repo

And then push to a branch in your fork.

And then create a pull request against this repository so we can review it.

On Sun, Apr 28, 2019 at 9:58 AM Abishek Ganesh [e-mail]
wrote:

@thunterdb https://github.com/thunterdb I am trying to push my code the
repo but I am not authorized to do so I guess. Kindly let me know how am I
supposed to push my changes to repo.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
https://github.com/databricks/koalas/issues/152#issuecomment-487397023,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AACO6PAWBUWKCPQJ24N65W3PSXJT3ANCNFSM4HH6733A
.

[abishekganesh72]

Thanks for the info. Let me do it that way.

[shril]

@rxin @thunterdb I would like to take this up as well.

[rxin]

@shril please do. We need all the help we can get. Thanks!

[ezwiefel]

@rxin @thunterdb Hope you're still looking for help. I'm going to submit a PR for some of the Series functions (starting with head).

[abishekganesh72]

@ezwirfel, Thanks for the help but I am working on the methods in series.py.

[ezwiefel]

@abishekganesh72, Sounds good - thanks!

[HyukjinKwon]

Go ahead guys!

[nitlev]

Hey, I started working on Dataframe.pipe if that's okay. I'll have to change a bit the docstring as the one used in pandas won't make the tests pass, but I think I can get something working soon.

[rxin]

@nitlev please do. Thanks for working on it.