.last() does not perform as expected

[sursu]

I find that .last() does not perform as expected.

Example:

df = pd.DataFrame([[179293473,'2016-06-01 00:00:03.549745','http://www.dr.dk/nyheder/',39169523],[179293473,'2016-06-01 00:04:22.346018','https://www.information.dk/indland/2016/05/hvert-tredje-offer-naar-anmelde-voldtaegt-tide', 39125224],
 [179773461, '2016-06-01 22:13:16.588146', 'https://www.google.dk', 31658124],
 [179773461, '2016-06-01 22:14:04.059781', 'https://www.google.dk', 31658124],
 [179773461, '2016-06-01 22:16:37.230587', np.nan, 31658124],
 [179773461, '2016-06-01 22:23:09.847149', 'https://www.google.dk', 32718401],
 [179773461, '2016-06-01 22:23:55.158929', np.nan, 32718401],
 [179773461, '2016-06-01 22:27:00.857224', np.nan, 32718401]],
columns=['SessionID', 'PageTime', 'ReferrerURL', 'PageID'])

Problem description

When I run:
df.groupby('SessionID').last()
I get:

SessionID	PageTime	ReferrerURL	PageID
179293473	2016-06-01 00:04:22.346018	https://www.information.dk/indland/2016/05/hve...	39125224
179773461	2016-06-01 22:27:00.857224	https://www.google.dk	32718401

Expected Output

When, in fact, I was expecting the same result as obtained from:
df.groupby('SessionID').nth(-1)

SessionID	PageID	PageTime	ReferrerURL
179293473	39125224	2016-06-01 00:04:22.346018	https://www.information.dk/indland/2016/05/hve...
179773461	32718401	2016-06-01 22:27:00.857224	NaN

---

And while we are at .nth(), why does it mix up my column order?

Output of pd.show_versions()

INSTALLED VERSIONS

commit: None
python: 3.6.4.final.0
python-bits: 64
OS: Windows
OS-release: 10
machine: AMD64
processor: Intel64 Family 6 Model 79 Stepping 1, GenuineIntel
byteorder: little
LC_ALL: None
LANG: None
LOCALE: None.None

pandas: 0.22.0
pytest: 3.3.2
pip: 9.0.3
setuptools: 38.4.0
Cython: 0.27.3
numpy: 1.14.0
scipy: 1.0.0
pyarrow: None
xarray: None
IPython: 6.2.1
sphinx: 1.6.6
patsy: 0.5.0
dateutil: 2.6.1
pytz: 2017.3
blosc: None
bottleneck: 1.2.1
tables: 3.4.2
numexpr: 2.6.4
feather: None
matplotlib: 2.1.2
openpyxl: 2.4.10
xlrd: 1.1.0
xlwt: 1.3.0
xlsxwriter: 1.0.2
lxml: 4.1.1
bs4: 4.6.0
html5lib: 1.0.1
sqlalchemy: 1.2.1
pymysql: None
psycopg2: 2.7.4 (dt dec pq3 ext lo64)
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None

[GriffinHines]

Working on this. Why not just have last() reference nth(-1)?

[sursu]

It seems that this issue has been around for quiet some time: here #6732, and here #8427.

@GriffinHines If nth(-1) performs as fast as the intended .last() why even have a .last() function?

[jreback]

duplicate of #8427

[sursu]

@jreback I see you have closed this issue, and not because it would be.. close to being closed.

#8427 exists for nearly 4 years.
As a user I am curious:

• Why hasn't it been resolved?
• Wouldn't it be a good idea there at least to be a warning in the documentation about this?

[jreback]

@sursu because its a duplicate, which is already open.

you are welcome to contribute a fix for either of these things. We have 2400 open issues. How would you prioritize things?

[sursu]

@jreback I understand.
But, as @GriffinHines suggested, a temporary and very quick solution could be just to reference .nth(-1).
Yes, it's patching, but I believe people prefer that to bugs.