GroupBy.nth includes group key inconsistently

[sinhrks]

Code Sample, a copy-pastable example if possible

nth doesn't inlcude group key as the same as first and last.

df = pd.DataFrame({'A': [1, 2, 3, 4, 5], 'B': [1, 2, 3, 4, 5],
                   'G': [1, 1, 2, 2, 1]})

g = df.groupby('G')
g.nth(1)
#    A  B
# G      
#1  2  2
#2  4  4

However, calling head makes the behavior change. Looks to be caused by _set_selection_from_grouper caches its selection.

g = df.groupby('G')
g.head()
g.nth(1)
#    A  B  G
# G         
#1  2  2  1
#2  4  4  2

Expected Output

always as below.

g.nth(1)
#    A  B
# G      
#1  2  2
#2  4  4

output of pd.show_versions()

current master.

[jreback]

this is fixed in #11039 (but it IS a change)

[jreback]

also some related issues there #7569

[sinhrks]

Thx, haven't noticed. This is also related to #7453 (last nth issue).

[sinhrks]

We can enable nth tests locates on test_resample after both this and #12840 are fixed.

[sinhrks]

This isn't fixed by #11039. Reopen.

[jreback]

In [1]: df = pd.DataFrame({'A': [1, 2, 3, 4, 5], 'B': [1, 2, 3, 4, 5],
   ...:                    'G': [1, 1, 2, 2, 1]})

In [2]: 

In [2]: g = df.groupby('G')

In [3]: g.nth(1)
Out[3]: 
   A  B
G      
1  2  2
2  4  4

In [4]: g.head()
Out[4]: 
   A  B
0  1  1
1  2  2
2  3  3
3  4  4
4  5  5

[sinhrks]

Call head before nth.

g = df.groupby('G')
g.head()
g.nth(1)
#    A  B  G
# G         
# 1  2  2  1
# 2  4  4  2

[jreback]

@sinhrks ahh, so something is changing state. ok!

[a-p-man]

On a related note, which version of g.head() below is expected?

In [2]: g = df.groupby('G')

In [3]: g.head()
Out[3]: 
   A  B  G
0  1  1  1
1  2  2  1
2  3  3  2
3  4  4  2
4  5  5  1

In [4]: g = df.groupby('G')

In [5]: g.nth(1)
Out[5]: 
   A  B
G      
1  2  2
2  4  4

In [6]: g.head()
Out[6]: 
   A  B
0  1  1
1  2  2
2  3  3
3  4  4
4  5  5