Implement DBSCAN as a Static transformer?

[ablaom]

using MLJ
M = @load DBSCAN
model = M()
X, _ = make_moons()
mach = machine(m, X)|> fit!

julia> transform(mach, X)
ERROR: MethodError: no method matching transform(::MLJScikitLearnInterface.DBSCAN, ::PyCall.PyObject, ::Tables.MatrixTable{Matrix{Float64}})
Closest candidates are:
  transform(::Union{MLJIteration.DeterministicIteratedModel{M}, MLJIteration.ProbabilisticIteratedModel{M}} where M, ::Any, ::Any) at /Users/anthony/.julia/packages/MLJIteration/Twn0E/src/core.jl:129
  transform(::Machine{var"#s471", C} where {var"#s471"<:Static, C}, ::Any, ::Any...) at /Users/anthony/.julia/packages/MLJBase/W4gFl/src/operations.jl:96
  transform(::Union{Composite, Surrogate}, ::Any, ::Any) at /Users/anthony/.julia/packages/MLJBase/W4gFl/src/operations.jl:122
  ...
Stacktrace:
 [1] transform(mach::Machine{MLJScikitLearnInterface.DBSCAN, true}, Xraw::Tables.MatrixTable{Matrix{Float64}})
   @ MLJBase ~/.julia/packages/MLJBase/W4gFl/src/operations.jl:88
 [2] top-level scope
   @ REPL[23]:1

Looks to me like transform is not implemented (confirmed by info(m)) and that there should be a line like the following one for the Birch transformer

MLJScikitLearnInterface.jl/src/models/clustering.jl

Line 68 in 0bc0bd5

@sku_transform Birch

in the code block defining BDSCAN (in the same file).

[ablaom]

Oops. I remember why this isn't implemented. DBSCAN only delivers cluster labels for training data. So this should be re-conceived as a Static transformer.

See discussion at JuliaAI/MLJClusteringInterface.jl#11, which will probably render this issue low priority.