BUG: pandas.groupby.apply does not work with floating point indexed dataframe since 1.0.0

[r-luo]

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• (optional) I have confirmed this bug exists on the master branch of pandas.

---

Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Code Sample, a copy-pastable example

import pandas as pd  # pandas==1.0.4, but this issue exists since 1.0.0
import numpy as np  # numpy==1.16.3

toy_df = pd.DataFrame(
    {
        'group': np.random.randint(0, 5, 10), 
        'value': np.random.random(10)
    }, 
    index=np.random.random(10),
)

# fails with floating point index
toy_df.groupby('group').apply(sum)
toy_df.groupby('group').apply(lambda x: x['value'].sum())

# with integer index then it works:
toy_df.index = np.random.randint(0, 5, 10)
toy_df.groupby('group').apply(lambda x: x['value'].sum())
toy_df.groupby('group').apply(sum)

Problem description

[this should explain why the current behaviour is a problem and why the expected output is a better solution]

I found this peculiar issue with the new version of pandas. When I have floating point index in a pandas DataFrame, then I cannot perform any groupby.apply operation on it. It will always give the below error:

ValueError: index must be monotonic increasing or decreasing
During handling of the above exception, another exception occurred:
...
pandas/_libs/index.pyx in pandas._libs.index.IndexEngine.get_loc()
pandas/_libs/index.pyx in pandas._libs.index.IndexEngine.get_loc()
pandas/_libs/hashtable_class_helper.pxi in pandas._libs.hashtable.Float64HashTable.get_item()
pandas/_libs/hashtable_class_helper.pxi in pandas._libs.hashtable.Float64HashTable.get_item()
KeyError: 0.0

When I used ipdb to trace inside the error, I found that the following line caused the problem:

> pandas/core/groupby/ops.py(930)fast_apply()
    929 
--> 930         sdata = self._get_sorted_data()
    931         return libreduction.apply_frame_axis0(sdata, f, names, starts, ends)

after this line, sdata still does not have a sorted index, and then the next line raises the ValueError: index must be monotonic increasing or decreasing.

I tested that this error exists in pandas==1.0.4 and pandas==1.0.0 but does not exist in pandas==0.25.1.

Expected Output

Expected output from groupby apply should be the same in the example with either floating point index or integer index.

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.7.0.final.0
python-bits : 64
OS : Linux
OS-release : 4.14.173-137.229.amzn2.x86_64
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8

pandas : 1.0.4
numpy : 1.16.3
pytz : 2020.1
dateutil : 2.8.0
pip : 20.0.2
setuptools : 47.1.1.post20200604
Cython : 0.29.2
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.10.1
IPython : 7.15.0
pandas_datareader: None
bs4 : None
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : None
matplotlib : 3.1.1
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 0.17.1
pytables : None
pytest : None
pyxlsb : None
s3fs : 0.4.0
scipy : 1.4.1
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
xlsxwriter : None
numba : None

[rhshadrach]

Thanks for reporting this. I think this may be a duplicate of #34455.

[r-luo]

@rhshadrach Yeah it seems to be the same underlying problem.

[rhshadrach]

Closing as duplicate of #34455