DataFrameGroupBy.ffill() and .bfill() remove column upon which it was grouped

[Koojav]

Code Sample, a copy-pastable example if possible

import pandas as pd
df = pd.DataFrame({'a': [1, 1, 1, 2], 'b': [11, 11, None, None]})

#   a     b
# 0  1  11.0
# 1  1  11.0
# 2  1   NaN
# 3  2   NaN

df.groupby('a').ffill()

#      b
# 0  11.0
# 1  11.0
# 2  11.0
# 3   NaN

Problem description

Filling missing values in groups removes the column upon which DataFrame got grouped by.
Previously (0.20.2) list of columns was left intact after such operation.

Expected Output

DataFrame containing the same list of columns as before the operation.

Output of pd.show_versions()

INSTALLED VERSIONS

commit : None
python : 3.6.7.final.0
python-bits : 64
OS : Linux
OS-release : 4.15.0-55-generic
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_US.UTF-8
LOCALE : en_US.UTF-8
pandas : 0.25.0
numpy : 1.16.4
pytz : 2019.1
dateutil : 2.8.0
pip : 19.2.1
setuptools : 40.6.3
Cython : 0.27.3
pytest : 3.0.7
hypothesis : None
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : 4.2.6
html5lib : None
pymysql : None
psycopg2 : 2.7.3.2 (dt dec pq3 ext lo64)
jinja2 : 2.10
IPython : 7.2.0
pandas_datareader: None
bs4 : 4.6.3
bottleneck : None
fastparquet : None
gcsfs : None
lxml.etree : 4.2.6
matplotlib : 2.0.2
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 0.13.0
pytables : None
s3fs : None
scipy : 1.0.0
sqlalchemy : 1.1.18
tables : None
xarray : None
xlrd : None
xlwt : None
xlsxwriter : None

[jreback]

see the whatsnew https://pandas.pydata.org/pandas-docs/stable/whatsnew/v0.25.0.html#dataframe-groupby-ffill-bfill-no-longer-return-group-labels

this was a regression

[Koojav]

@jreback the behavior is not consistent at all. For .sum() grouped column values are moved to the index series while .ffill() just gets rid of them.

Proof:

df = pd.DataFrame({'a': ['a','b','c'], 'v':[1,2,3]})

df.groupby('a').sum()
Out[21]: 
   v
a   
a  1
b  2
c  3

df.groupby('a').ffill()
Out[22]: 
   v
0  1
1  2
2  3

[jreback]

https://pandas.pydata.org/pandas-docs/stable/user_guide/groupby.html#aggregation

sum is an aggregation while filling is a transformation

[chrisjcameron]

@jreback: Is the note at the end of the transformation section no longer valid?

https://pandas.pydata.org/pandas-docs/stable/user_guide/groupby.html#transformation

Note
Some functions will automatically transform the input when applied to a GroupBy object, but returning an object of the same shape as the original. Passing as_index=False will not affect these transformation methods.
For example: fillna, ffill, bfill, shift..

Is there a chained/functional syntax to fill missing values? Is
df.update(df.sort_values(['key1', 'time']).groupby('key1').ffill()) the only option to get a dataframe with the na's filled in?

[AnotherSamWilson]

For anyone else that found this annoying, I found that this works best:

key_col = df['key'].copy()
df.groupby(key_col).ffill()

For multiple columns, you can do:

key_col1 = df['key1'].copy()
key_col2 = df['key2'].copy()
df.groupby([key_col1, key_col2]).ffill()