Code Clean Up - groupby_helper.pxi.in

[WillAyd]

A few of the methods in groupby_helper.pxi.in have a pattern that goes:

    N, K = (<object> values).shape

    with nogil:

        if K > 1:
            # loop over j in range(K) and do stuff
            ...
        else:
            # do stuff with j=0
            ...

Unless I'm missing something simple I don't think there's a reason to have the conditional and duplicate code branches. In the case where K=1 (which is all the time?) then the for j in range(K) construct inherently sets j=0, bypassing the need for the conditional

[chris-b1]

I believe the K > 1 would get used sometime, from what I remember some of the aggregations are done by-block, which means a 2d array of multiple columns could get passed down.

pandas/pandas/core/groupby.py

Line 3736 in 2fdf1e2

def _cython_agg_general(self, how, alt=None, numeric_only=True,

It's probably a micro-micro optimization, but having it split like this may generate slightly better code for K=1 case. But welcome you to benchmark and if it doesn't matter, clean it up.