Improve querying for measures

Open ablaom opened this issue 4 years ago • 1 comments

There used to be measures(matching(y)) but that didn't really make sense because y doesn't tell you whether you want to running a probabilistic model or a deterministic one. @f0lie suggests basing the query on a ready-constructed machine, which makes more sense. See https://github.com/alan-turing-institute/MLJ.jl/issues/753#issuecomment-805378211 .

Maybe something like measures(matching(mach))??

Mar 25 '21 05:03 ablaom

The problem with this, of course, is that measures should ideally be conceptualised as independent of MLJ altogether, and machines are an MLJ concept.

Apr 08 '21 23:04 ablaom