CLN/DEPR: Consistent naming for isnull, notnull, fillna, dropna

[jpeacock29]

In 2.0, these functions could be named more consistently, either all *na or *null. Or, to preserve backwards compatibility, isna, notna, fillnull and dropnull could be added as aliases. Either way, it would be helpful to take the guess work out of this.

Let me know if this is not the correct forum for such proposals.

[jreback]

these names have been ingrained for quite a long time, so changing is painful w/o much gain. That said, I suppose these could be slightly confusing.

I think adding aliases would be ok. Deprecating and using fillnull, dropnull would be ok too (though again not sure of the gain here)

@shoyer @TomAugspurger @sinhrks @TomAugspurger @jorisvandenbossche @wesm

[jorisvandenbossche]

If we add aliases, I would personally rather go for the *na ones instead of the adding the *null ones. Certainly if we are going to use NA as the universal missing value indicator for the different data types.

[wesm]

+1 to isna and notna. The null variants are quite ingrained -- I am neutral on deprecating/removing them

[sinhrks]

+1 for na, as some methods has args named na_xxx.