Description
Code Sample, a copy-pastable example if possible
import pandas as pd
data = pd.DataFrame(dict(values=range(20), quintiles=pd.cut(range(20), 5)))
grouped = data.groupby("quintiles")["values"]
samples = grouped.agg(lambda x: list(x.sample(3)))
samples
# quintiles
# (-0.019, 3.8] [3, 2, 1]
# (3.8, 7.6] [7, 4, 6]
# (7.6, 11.4] [8, 10, 11]
# (11.4, 15.2] [15, 14, 12]
# (15.2, 19.0] [17, 16, 19]
# Name: values, dtype: object
samples[0]
# [3, 2, 1]
samples[1]
# [3, 2, 1]
samples[2.3]
# [3, 2, 1]
samples[19]
# [17, 16, 19]
samples.keys()
# CategoricalIndex([(-0.019, 3.8], (3.8, 7.6], (7.6, 11.4], (11.4, 15.2],
# (15.2, 19.0]],
# categories=[(-0.019, 3.8], (3.8, 7.6], (7.6, 11.4], (11.4, 15.2], (15.2, 19.0]], ordered=True, name='quintiles', dtype='category')Problem description
Indexing a series object that has pd.cut() intervals as its index results in membership test of the provided index on the intervals, which I found interesting, but it was very unexpected. Is this intentional and documented somewhere? When I give the labels names (labels=list("abcde") in pd.cut), this doesn't happen anymore.
Expected Output
I expectived it to fall back on .iloc[].
Output of pd.show_versions()
INSTALLED VERSIONS
commit: None
python: 3.7.3.final.0
python-bits: 64
OS: Linux
OS-release: 3.10.0-957.12.2.el7.x86_64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: None
LANG: en_GB.UTF-8
LOCALE: en_GB.UTF-8
pandas: 0.24.2
pytest: 4.4.1
pip: 19.1.1
setuptools: 41.0.1
Cython: 0.29.6
numpy: 1.16.3
scipy: 1.2.1
pyarrow: None
xarray: 0.12.1
IPython: 7.4.0
sphinx: None
patsy: 0.5.1
dateutil: 2.8.0
pytz: 2019.1
blosc: None
bottleneck: None
tables: None
numexpr: None
feather: None
matplotlib: 3.1.0
openpyxl: None
xlrd: None
xlwt: None
xlsxwriter: None
lxml.etree: 4.3.3
bs4: 4.7.1
html5lib: None
sqlalchemy: None
pymysql: None
psycopg2: None
jinja2: 2.10
s3fs: None
fastparquet: None
pandas_gbq: None
pandas_datareader: None
gcsfs: None