BUG: Inconsistent behaviour for DataFrame.value_counts and Series.value_counts on categoricals

[corriebar]

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

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

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

Reproducible Example

import pandas as pd

df = (pd.DataFrame({'gender' : ['male', 'female', 'male', 'female', 'male'],
                  'country': ['BR', 'BR', 'US', 'BR', 'US']})
      .astype('category') 
     )
df.value_counts() # drops all empty categories
# gender  country
# female  BR         2
# male    US         2
#         BR         1
# dtype: int64

# also compare
(df
 .query("country != 'BR'")
 .value_counts()
) # drops non-observed categories
# gender  country
# male    US         2
# dtype: int64

(df
 .query("country != 'BR'")
 ['gender']
 .value_counts()
) # shows 0 for non-observed catgories
# male      2
# female    0
# Name: gender, dtype: int64

Issue Description

DataFrame.value_counts() drops any non observed categories while Series.value_counts() shows 0 for non-observed categories.

There was some discussion about the zeroes for Series.value_counts() in this issue #8559, based on that I checked behaviour in R which also keeps the zeros:

  gender <- factor(c('male', 'female', 'male', 'female', 'male'))
  country <- factor( c( 'BR', 'BR', 'US', 'BR', 'US') )
df <- data.frame(gender=gender, country=country)
table(df)
#>         country
#> gender   BR US
#>   female  2  0
#>   male    1  2

Expected Behavior

The following work-around gives the expected behaviour:

df.groupby("gender").country.value_counts()
# gender    
# female  BR    2
#         US    0
# male    US    2
#         BR    1
# Name: country, dtype: int64

This fails however for more than two categorical variables.

Installed Versions

INSTALLED VERSIONS

commit : 73c6825
python : 3.9.7.final.0
python-bits : 64
OS : Darwin
OS-release : 19.6.0
Version : Darwin Kernel Version 19.6.0: Thu Sep 16 20:58:47 PDT 2021; root:xnu-6153.141.40.1~1/RELEASE_X86_64
machine : x86_64
processor : i386
byteorder : little
LC_ALL : None
LANG : None
LOCALE : None.UTF-8

pandas : 1.3.3
numpy : 1.21.2
pytz : 2021.1
dateutil : 2.8.2
pip : 21.2.4
setuptools : 58.0.4
Cython : None
pytest : None
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : None
IPython : 7.27.0
pandas_datareader: None
bs4 : None
bottleneck : None
fsspec : None
fastparquet : None
gcsfs : None
matplotlib : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlwt : None
numba : None

[NumberPiOso]

take

[NumberPiOso]

We were following an outdated issue. The expected behavior changed dramatically.
PR #20583 API: categorical grouping will no longer return the cartesian product
Due to some BUGs as in #14942.

With that said, I think this issue should be closed.

[rhshadrach]

The title to #20583 is no longer accurate; that PR ended up adding the observed keyword to groupby instead. This issue is still relevant.