ENH: GroupBy.transform should accept similar arguments to GroupBy.agg

[WillAyd]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

In [8]: df = pd.DataFrame({"col": list("aab"), "val": range(3)})
In [8]: df.groupby("col").agg(["sum", "min"])
Out[8]: 
    val    
    sum min
col        
a     1   0
b     2   2
In [9]: df.groupby("col").transform(["sum", "min"])
TypeError: 'list' object is not callable

Feature Description

To get people out of the mindset of always using apply, it would be nice if .transform supported the same call signature as .agg

Additionally it would be great to support NamedAgg somehow in transform. The naming is a bit unfortunate...we could create a NamedTransform alias or maybe just live with the API inconsistency of:

.agg(col_a=NamedAgg("col", "sum"), ...)
.transform(col_a=NamedAgg("col", "cumsum"), ...)

I think either is still a net improvement over today

Alternative Solutions

n/a

Additional Context

No response

[abeltavares]

take

[rhshadrach]

I'm positive here, but there are potentially issues with agg and list-likes and we need to decide if we want to align to agg's behavior, or diverge from it. Namely, agg with a list-like will break up a DataFrame into Series first (even with a string like "sum"), and then operate on the Series.

Ref:

• BUG: Can't pass arguments to DataFrameGroupBy.agg when using list/dict-like and string aliases #53839
• PERF: Calling df.agg([function]) is much slower than df.agg(function) when there are many columns and few rows. #45658

[abeltavares]

@WillAyd @rhshadrach Hi guys, the GroupBy.transform repeat aggregate values for same number of rows like original DataFrame, as opposite to GroupBy.agg, that behaviour is to be kept right?

[abeltavares]

@WillAyd @rhshadrach Hi guys, the GroupBy.transform repeat aggregate values for same number of rows like original DataFrame, as opposite to GroupBy.agg, that behaviour is to be kept right?

@WillAyd @rhshadrach can you confirm that?

[rhshadrach]

@abeltavares - that is correct.