Series.sort_values doesn't support a `by` keyword argument, but the documentation claims it does

[mineo]

pandas: 0.17.0

import pandas
s = pandas.Series([1,2,3])
s.sort_values(by=["a"])

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-4-4a7ccaa074a5> in <module>()
      1 import pandas
      2 s = pandas.Series([1,2,3])
----> 3 s.sort_values(by=["a"])

TypeError: sort_values() got an unexpected keyword argument 'by'

https://github.com/pydata/pandas/blob/1135ce34fc2fb995e239a7d36aee1b687ef69dd9/pandas/core/series.py#L1609 does indeed not support by, but the docs list it as a supported keyword argument.

[jreback]

this is just using the super docs
want to do a pr to fix?

[mineo]

Frame.sort_values and NDFrame.sort_values support by, though, so I'm not sure how to fix them.

[jreback]

take the by out of the shared docs and then add back as part of the doc string for DataFrame and NDFrame

[mineo]

Important things first: I still think pandas is a really cool thing ✨, but I give up :( Building the docs without conda is just too hard - installing everything from ci/requirements_dev.txt doesn't even get you into a state where python make.py html actually works, I had to install matplotlib manually and run python setup.py build_ext --inplace. The generated docs are now missing large parts of the API reference (including Series.sort_values). This might be fixed by installing everything in ci/requirements_all.txt but the package names in there differ from the package names on pypi and I don't have conda installed. I'm sorry, but I currently do not have the time to figure out how to use another tool for building python environments just to remove this one line of documentation, so don't expect a pull request from me.

[jreback]

@mineo

this is a doc-string! not the actual docs, you don't actually need to build the docs.

you can simply examine them from ipython: Series.sort_values?

though it IS pretty easy to build the docs (but you should for sure use conda if you do), see guidelines here

[jreback]

and you need cannot JUST install requirements_dev.txt that is just a minimal set. The docs are quite comprehensive on these points; you need quite a lot of deps to build the docs.

[varunkumar-dev]

@jreback If I take the by out from the shared docs and then add "by" parameter as part of the doc string for DataFrame and NDFrame, it will not include "by" in the parameter section. It is showing "by" docstring on top of shared doc .
In that case , should I remove sort_values from shared docs and add different doc strings in Frame.sort_values and Series.sort_values ?

[pau1a]

I just came up against this tonight. I this just lingering? Is the documentation wrong or is this a bug?

[jorisvandenbossche]

This should actually already be fixed by #13496 (released in 0.19.0) in the meantime.
@pau1a what version of pandas are you using?

[iamlydialawal]

Yeah this was not fixed...might coerce back into a df