DOC: df.groupby('A') is just syntactic sugar for df.groupby(df['A'])

[Sirmadeira]

Pandas version checks

• I have checked that the issue still exists on the latest versions of the docs on main here

Location of the documentation

https://pandas.pydata.org/docs/user_guide/groupby.html

Especifically on the line

df.groupby('A') is just syntactic sugar for df.groupby(df['A']).

A list of any of the above things.

Documentation problem

Well here is a sample on how is not just syntactic sugar. I think

test_df = pd.DataFrame({'Category': {0: 'product-availability address-confirmation input',
  1: 'registration register-data-confirmation options',
  2: 'onboarding return-start input',
  3: 'registration register-data-confirmation input',
  4: 'decision-tree first-interaction-validation options'},
 'Original_UserId': {0: '5511949551865@wa.gw.msging.net',
  1: '5511949551865@wa.gw.msging.net',
  2: '5511949551865@wa.gw.msging.net',
  3: '5511949551865@wa.gw.msging.net',
  4: '5511949551865@wa.gw.msging.net'}})

If I run
test_df['Category'].eq('onboarding return-start input').groupby(test_df['Original_UserId']).cummax()

This gives a result

If I run

test_df['Category'].eq('onboarding return-start input').groupby('Original_UserId').cummax()
I get keyerror

I am guessing the keyerror is because of the checking that occurs on the given object, that being whether the object contains that given column or not.

Suggested fix for documentation

I am not sure, maybe just add that the difference is that one checks whether one contains the given object series and the other does not.

[phofl]

cc @jorisvandenbossche

Did we figure this out completely for CoW? Otherwise would just remove this line

[rhshadrach]

@Sirmadeira

df.groupby('A') is just syntactic sugar for df.groupby(df['A']).

In the statement, the df inside and outside the groupby must be the same object; this is violated in the example you provided.

[Sirmadeira]

True they are different objects.

[rhshadrach]

@phofl - you good with this closed?

[phofl]

yeah the CoW topic is already known and tracked, so no need to keep open here

[jorisvandenbossche]

Did we figure this out completely for CoW? Otherwise would just remove this line

For future reference, the PR for CoW on that is further looking at this is #50730 (which has the references to the original discussion).

The bottom line is that the "syntactic sugar" is a bit more costly to ensure for CoW