DOC: undocumented head(n), tail(n) accept negative values. But not on GroupBy

[smcinerney]

DOC: undocumented behavior that works: head(n), tail(n)accept negative values both for DataFrame, Series. But not for GroupBy.head(n) ; doesn't correctly handle negative values.

This is quite useful and could be documented.

• head(-n) returns all rows(/elements) other than the n first values. Similar to df[-6::].
• tail(-n) returns all rows(/elements) other than the n last values
• There is no error if the length n specified is too long for the dataframe(/series), just silently returns an empty result.

Code Sample, a copy-pastable example if possible

df = pd.DataFrame({'x': range(18,0,-1), 'y': [x % 6 for x in range(18)] })

df.head(-25)
Empty DataFrame
Columns: [x, y]
Index: []

df.head(-15)
    x  y
0  18  0
1  17  1
2  16  2

 df.tail(-15)
    x  y
15  3  3
16  2  4
17  1  5

df['x'].tail(-15)
15    3
16    2
17    1

# But GroupBy.head(n) doesn't
 df.groupby('y').head(-1)
Empty DataFrame
Columns: [x, y]
Index: []
df.groupby('y').tail(-1)
Empty DataFrame
Columns: [x, y]
Index: []

Problem description

Add to documented behavior.

Expected Output

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.7.3.final.0
python-bits: 64
OS: Darwin
OS-release: 18.7.0
machine: x86_64
processor: i386
byteorder: little
LC_ALL: None
LANG: en_US.UTF-8
LOCALE: en_US.UTF-8

pandas: 0.24.2
pytest: 5.2.4
pip: 19.3.1
setuptools: 41.6.0.post20191030
Cython: 0.29.14
numpy: 1.17.3
scipy: 1.3.1
pyarrow: None
xarray: None
IPython: 7.9.0
sphinx: 2.2.1
patsy: 0.5.1
dateutil: 2.8.1
pytz: 2019.3
blosc: None
bottleneck: 1.3.1
tables: 3.6.1
numexpr: 2.7.0
feather: None
matplotlib: 3.1.1
openpyxl: 3.0.1
xlrd: 1.2.0
xlwt: 1.3.0
xlsxwriter: 1.2.6
lxml.etree: 4.4.1
bs4: 4.8.1
html5lib: 1.0.1
sqlalchemy: 1.3.11
pymysql: None
psycopg2: 2.8.4 (dt dec pq3 ext lo64)
jinja2: 2.10.3
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None
gcsfs: None

[smcinerney]

Subsequently found out this might be a duplicate of #9214, except its title and body don't make clear that it's suggesting the undocumented behavior is useful and should be documented.

[TomAugspurger]

Happy to have this documented / implemented on groupby. Are you interested in working on it?

[smcinerney]

This is marked as a docbug; I was mainly interested in documenting (both in docstring and pandas-docs) that this works on DataFrame, Series; I'm happy to do that if you can tell me how. Don't care much about implementing it on groupby.

[MarcoGorelli]

Seems that this was closed by #30556, feel free to reopen if you disagree