Anthony Blaom, PhD

@bkamins Thanks indeed for your thoughts. I didn't quite understand this comment: > In general e.g. a vector of NamedTuple is a table, not only a Tuple of them: The...

@bkamins Thanks for clarifying the comment! > Given your comments maybe you want to propose another rule: > Never treat Tuple to be a table and make Tables.istable return false...

> Why not just add that restriction in ScientificTypes only? @nalimilan Thanks for looking into this. > What would be the rationale for this in Tables.jl? The rational is I...

@nalimilan Thanks. I will suggest, then, that those two packages treat all tuples as non-tabular objects, and document the fact.

@quinnj 🙏🏾

Mmm. I'm not super happy with removing "rows" from the the name. If I'm new to tables in julia, "subset" could mean anything. I similarly never liked name like "select"...

Yes, I get that it is not returning "rows" in the sense of not returning a row iterator. But I'm not sure general users equate "rows" with "row iterator", maybe...

Names are important. I struggled for some hours to make sense of the MLR documentation because I could not undersand what the `trafo` method was supposed to be. Once I...

> Support non one-hot encoded categorical features: features carrying item info as an Int (1 to N levels). Maybe better to support categorical arrays to deal with classes that disappear...

Yes, we have both. See https://alan-turing-institute.github.io/MLJ.jl/dev/adding_models_for_general_use/#Document-strings . If you also have a "native" interface for your MLJ objects, this can of course be worked into the doc string. Happy to...