Group by apply
During our discussion on algorithms and how to apply them back in Chapter 5, Algorithms and How to Apply Them, we came across the Apply function, which is both powerful and terrifying at the same time. An equivalent function for group by exists as pd.core.groupby.DataFrameGroupBy.apply with all of the same caveats. Generally, this function is overused, and you should opt for pd.core.groupby.DataFrameGroupBy.agg or pd.core.groupby.DataFrameGroupBy.transform instead. However, for the cases where you don’t really want an aggregation or a transformation, but something in between, using apply is your only option.
Generally, pd.core.groupby.DataFrameGroupBy.apply should only be used as a last resort. It can produce sometimes ambiguous behavior and is rather prone to breakage across releases of pandas.
How to do it
In the There’s more… section of the previous recipe, we mentioned how it is not possible to start with a pd.DataFrame of the following...
