BUG: groupby apply inconsistent output

[gurukiran07]

• 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.

---

Code Sample, a copy-pastable example

import pandas as pd
import numpy as np

df1 = pd.DataFrame({'group_col': [0.0, np.nan, 0.0, 0.0], 'value_col': [2,2,2,2]})
df2 = pd.DataFrame({'group_col': [np.nan, 0.0, 0.0, 0.0], 'value_col': [2,2,2,2]})

groupby apply on df1 gives multi-index output

df1.groupby('group_col').value_col.apply(lambda x: x.value_counts().reindex(index=[1,2,3]))

group_col   
0.0        1    NaN
           2    3.0
           3    NaN
Name: value_col, dtype: float64

groupby apply on df2 gives a single level index output:

df2.groupby('group_col').value_col.apply(lambda x: x.value_counts().reindex(index=[1,2,3]))

0    NaN
1    NaN
2    3.0
3    NaN
Name: value_col, dtype: float64

Problem description

groupby apply on df1 and df2 produce different outputs. They should produce the same output.

Expected Output

The output of both the above mentioned examples should be this:

group_col   
0.0        1    NaN
           2    3.0
           3    NaN
Name: value_col, dtype: float64

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.7.4.final.0
python-bits : 64
OS : Windows
OS-release : 10
machine : AMD64
processor : Intel64 Family 6 Model 94 Stepping 3, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : None.None

pandas : 1.0.3
numpy : 1.18.3
pytz : 2019.3
dateutil : 2.8.1
pip : 20.1.1
setuptools : 46.1.3
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.13.0
pandas_datareader: None
bs4 : None
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : None
matplotlib : 3.2.1
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pytables : None
pytest : None
pyxlsb : None
s3fs : None
scipy : 1.4.1
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : 1.2.0
xlwt : None
xlsxwriter : None
numba : None

[jendrikjoe]

Not sure, if it is the exact same issue, but we see a similar inconsistency when doing conditional filtering. A minimal reproducible example:

df = pd.DataFrame(
    index=["a", "b", "c", "d"], 
    data=[(0, 1.), (0, 2.), (1, 3.), (1, 4.)],
    columns=["group_col", "value"]
)

Doing an apply with a value that actually triggers filtering (<3):

df.groupby("group_col").apply(
    lambda df: df[["value"]] if all(df["value"] < 3) else df.iloc[1:][["value"]]
)

>>>           value
group_col
0         a    1.0
          b    2.0
1         d    4.0

Doing an apply with a value that does not trigger filtering (<5):

df.groupby("group_col").apply(
   lambda df: df[["value"]] if all(df["value"] < 5) else df.iloc[1:][["value"]]
)

>>>  value
a    1.0
b    2.0
c    3.0
d    4.0

Expected output:

I would expect the latter to output:

               value
group_col
0         a    1.0
          b    2.0
1         c    3.0
          d    4.0

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit : None python : 3.8.2.final.0 python-bits : 64 OS : Darwin OS-release : 19.4.0 machine : x86_64 processor : i386 byteorder : little LC_ALL : en_US.UTF-8 LANG : en_US.UTF-8 LOCALE : en_US.UTF-8

pandas : 1.0.3
numpy : 1.18.4
pytz : 2020.1
dateutil : 2.8.1
pip : 20.0.2
setuptools : 46.1.3
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.14.0
pandas_datareader: None
bs4 : None
bottleneck : None
fastparquet : 0.4.0
gcsfs : None
lxml.etree : None
matplotlib : 3.2.1
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pytables : None
pytest : None
pyxlsb : None
s3fs : None
scipy : 1.4.1
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
xlsxwriter : None
numba : 0.49.1

[gurukiran07]

@jendrikjoe Might be related. Not sure.

Example:

df2.groupby('group_col').value_col.apply(lambda x: x.value_counts().reindex(index=[1,2,3]))

0    NaN
1    NaN
2    3.0
3    NaN
Name: value_col, dtype: float64

df2.head(3).groupby('group_col').value_col.apply(lambda x: x.value_counts().reindex(index=[1,2,3]))

group_col   
0.0        1    NaN
           2    2.0
           3    NaN
Name: value_col, dtype: float64

[gimseng]

I played around with a little bit more with df1 and df2. If I changed the index list in the reindex argument, I get (the right multilevel answer):

df2.groupby('group_col').value_col.apply(lambda x: x.value_counts().reindex(index=[1,2]))

group_col   
0.0        1    NaN
           2    3.0
Name: value_col, dtype: float64

As another example, I shortened the data and try with different index list. Here are the few outputs:

df3 = pd.DataFrame({'group_col': [np.nan, 0.0,0.0], 'value_col': [2,2,2]})
df3.groupby('group_col').value_col.apply(lambda x: x.value_counts().reindex(index=[1,2]))

0    NaN
1    NaN
2    2.0
Name: value_col, dtype: float64

and

df4 = pd.DataFrame({'group_col': [np.nan, 0.0], 'value_col': [2,2]})
df4.groupby('group_col').value_col.apply(lambda x: x.value_counts().reindex(index=[1]))

0   NaN
1   NaN
Name: value_col, dtype: float64

I suspect there is something about (re)-indexing and missing value. I would love to work on this, but I am new to this repo. If someone else could point me towards the right direction, I'd be happy to investigate further.

Output of pd.show_versions()

Details

INSTALLED VERSIONS

commit : None
python : 3.6.9.final.0
python-bits : 64
OS : Linux
OS-release : 4.19.104+
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 : 1.18.4
pytz : 2018.9
dateutil : 2.8.1
pip : 19.3.1
setuptools : 47.1.1
Cython : 0.29.19
pytest : 3.6.4
hypothesis : None
sphinx : 1.8.5
blosc : None
feather : 0.4.1
xlsxwriter : None
lxml.etree : 4.2.6
html5lib : 1.0.1
pymysql : None
psycopg2 : 2.7.6.1 (dt dec pq3 ext lo64)
jinja2 : 2.11.2
IPython : 5.5.0
pandas_datareader: 0.8.1
bs4 : 4.6.3
bottleneck : 1.3.2
fastparquet : None
gcsfs : None
lxml.etree : 4.2.6
matplotlib : 3.2.1
numexpr : 2.7.1
odfpy : None
openpyxl : 2.5.9
pandas_gbq : 0.11.0
pyarrow : 0.14.1
pytables : None
pytest : 3.6.4
pyxlsb : None
s3fs : 0.4.2
scipy : 1.4.1
sqlalchemy : 1.3.17
tables : 3.4.4
tabulate : 0.8.7
xarray : 0.15.1
xlrd : 1.1.0
xlwt : 1.3.0
xlsxwriter : None
numba : 0.48.0

[rhshadrach]

I now get the expected output in the OP.

[parthi-siva]

take