Open tlienart opened 4 years ago
I guess you mean adding docs for subtyping Static
(which supersedes this StaticTransformer
).
There is some documentation, buried under "Composing Models" section of manual (search for Averager
but agree this is not the obvious place to find it. Perhaps under "Simple User Defined Models" with links from "Adding Models for General Use", and the existing entry.
Basically written, no?
https://github.com/alan-turing-institute/MLJ.jl/issues/307#issuecomment-594897959
Maybe change the function used example to simplify / improve readability? Maybe drop use of DataFrames to make more generic?
cc @vollmersj
There is now an entry in the user manual that also could be used.
https://alan-turing-institute.github.io/MLJ.jl/dev/transformers/#Static-transformers-1
On reconsideration, we should just use a standard pipeline, and not a learning network as these are more advanced.
one example given on slack would be recoding:
x -> replace(uppercase(x), " "=>"")