Description

I recently came across the need to filter groups by some condition and was surprised to find that polars does not natively support this.

Currently, one has to do something like the following to filter for groups that satisfy a particular property:

df
  .group_by("col")
  .agg(at_least_two=pl.len() >= 2)
  .filter("at_least_two")
  .drop("at_least_two")

This would be much more ergonomic if there was a SQL-like having construct:

df
  .group_by("col")
  .having(pl.len() >= 2)

While I personally like having as it distinguishes the context from the filter method, pandas calls such a method filter (docs).

Would you be open to adding something like this to polars?