ENH: union_categorical supports identical categories with ordered

[sinhrks]

• related to ENH: Categorical.from_union #13410, Appending Pandas dataframes in for loop results in ValueError #13524
• tests added / passed
• passes git diff upstream/master | flake8 --diff
• whatsnew not needed

.append / concat can handle ordered Categorical which has the identical categories. However, union_categorical raises TypeError. This PR allows union_categorical to handle the case.

on current mastser

s1 = pd.Series(pd.Categorical([1, 2, 3], ordered=True))
s2 = pd.Series(pd.Categorical([3, 2, 1], categories=[1, 2, 3], ordered=True))
s1.append(s2)

#0    1
#1    2
#2    3
#0    3
#1    2
#2    1
# dtype: category
# Categories (3, int64): [1 < 2 < 3]

pd.types.concat.union_categoricals([s1.values, s2.values])
# TypeError: Can only combine unordered Categoricals

CC: @JanSchulz, @chris-b1

[codecov-io]

Current coverage is 85.33% (diff: 100%)

Merging #13763 into master will increase coverage by <.01%

@@             master     #13763   diff @@
==========================================
  Files           141        141          
  Lines         50703      50717    +14   
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
+ Hits          43267      43281    +14   
  Misses         7436       7436          
  Partials          0          0          

Powered by Codecov. Last update 1367945...9cadc4e

[chris-b1]

has -> have

[chris-b1]

Looks good to me, other than a couple nit-picky doc things.

[chris-b1]

I created - sinhrks#1 to also add a sort_categories argument, which is needed for #13406

[jankatins]

In my thinking union_categorical is a "unstricter" (and "uncleaner") variant of the normal categorical handling: where the usual cat handling errors, this should "bend the rules" and succeed. Following this, I thinking union_categorical should handle whatever append can take...

[jankatins]

IMO this should read:

In addition to the "easy" case of combining two categoricals of the same categories and order information (e.g. what you could also append for), union_categoricals only works with unordered categoricals and will raise if any are ordered.

I.e. make it clear that union_categoricals supports something more than append.

[jreback]

@chris-b1 , @JanSchulz any final comments?

[chris-b1]

No, looks good to me.

On Jul 28, 2016 7:32 PM, "Jeff Reback" notifications@github.com wrote:

@chris-b1 https://github.com/chris-b1 , @JanSchulz
https://github.com/janschulz any final comments?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#13763 (comment), or mute
the thread
https://github.com/notifications/unsubscribe-auth/AB1b_FOSzzp-Gnvwk0SvEXA7FwPCj4fkks5qaUolgaJpZM4JTSrP
.

[jankatins]

LGTM

[jreback]

ty sir!