CLN: .value -> ._values outside of core/

[jbrockmendel]

pandas.plotting I left alone

[WillAyd]

lgtm

[jorisvandenbossche]

In general: do we just want to always use _values instead of values where possible? Or do we still prefer values in case both are not different (eg because of dealing with numeric data), in the idea of preferring a public property above a private, even in our own code base?

[jreback]

+1 on this, I agree we should not use .values at all; agree that using np.asarray() when we actually need a numpy arra (or .to_numpy()) explicitly is nice too. In an ideal world, we never should use .values and we can then maybe deprecate on the user side.

[jorisvandenbossche]

@jbrockmendel I am fine with the general rule of not using .values. Although I think in cases where we explicitly need a numpy array, we should rather use asarray / to_numpy.

But there are still some relevant inline comments

[jbrockmendel]

@jorisvandenbossche updated the one clearly-actionable inline comment. can you point out which other comments you have in mind? (or "resolve conversation" for the complement)

[jorisvandenbossche]

Added new comments to those that are IMO still actionable, and resolved the others

[jreback]

thanks