Make `DataFrame.pivot` accepts a list of column names as values

[dclong]

Problem description

Currently DataFrame.pivot accepts either None or a column name for the option values but not a list of column names.

# this works
df.pivot(values = None)
df.pivot(values = 'column_name')

# this doesn't work
df.pivot(values = ['cn1', 'cn2'])

It is not convenient if I want to create a pivot table for multiple columns at the same time. Of course, we can pass None to values and then select columns we want to. But this is not as convenient.

[gfyoung]

@dclong : Thanks for the issue! Have you tried "df.pivot_table" ? I think this might better address your needs. df.pivot is not meant for what you've described AFAIK.

[dclong]

@gfyoung: It seems that pandas.pivot_table requires an aggregation function? I don't want to aggregate the data but just pivot the table.

[gfyoung]

@dclong : Ah, gotcha. I'll tag this as an enhancement for now, though then again I'm not sure that that's how pivot is supposed to be used (or whether the original designers intended that to be allowed).

[jorisvandenbossche]

I think this is sensible to allow. pivot already either pivots a single column are all remaining columns, so I don't see a reason to not allow to pivot a selection of the columns (certainly because pivot_table also allows this).

PRs welcome!

[ibrahimsharaf]

Hello @jorisvandenbossche, I would love to work on this, how can I start?

[jorisvandenbossche]

@ibrahimsharaf Did you already go through the contributing guide (to set up development environment and such: http://pandas.pydata.org/pandas-docs/stable/contributing.html). If not, you should start with that.

For this issue: I would recommend to first write some tests for the desired new functionality, and then you can look at changing the pivot function to do this (the function is located here: the pivot function in this file: https://github.com/pandas-dev/pandas/blob/master/pandas/core/reshape/reshape.py)