Shouldn't HHI be calculated on masked p_records ?

[lcrmorin]

I was implementing and comparing some metrics.

In the calculation of the HHI here shouldn't this be masked - hhi(p_records[mask]) - to avoid including empty bins in the calculation ?

Best,
Lucas

[lcrmorin]

Similarly ks test seems to be calculated on raw proportions (p_event, p_nonevent), while other tests are calculated on masked versions (p_ev, p_nev).

[guillermo-navas-palencia]

Hi @lcrmorin. It seems sensible, I will have a look while working on the 1.0.0 version. Thanks