MLJ port WIP v2

Author: PyDataBlog

src/kmeans.jl

@@ -173,7 +173,7 @@ end
 
 
 """
-    Kmeans!(alg::AbstractKMeansAlg, containers, design_matrix, k; n_threads = nthreads(), k_init="k-means++", max_iters=300, tol=1e-6, verbose=true)
+    Kmeans!(alg::AbstractKMeansAlg, containers, design_matrix, k; n_threads = nthreads(), k_init="k-means++", max_iters=300, tol=1e-6, verbose=false)
 
 Mutable version of `kmeans` function. Definition of arguments and results can be
 found in `kmeans`.

src/mlj_interface.jl

@@ -6,39 +6,79 @@ using ParallelKMeans
 ####
 #### MODEL DEFINITION
 ####
-# TODO 2: MLJ-compatible model types and constructors,
+# TODO 2: MLJ-compatible model types and constructors
 @mlj_model mutable struct ParaKMeans <: MLJModelInterface.Unsupervised
     # Hyperparameters of the model
-    algo::AbstractKMeansAlg     = Lloyd::(_ in (Lloyd, Hamerly, Elkan))
-    k_init::String              = "k-means++"::(_ in ("k-means++", String))
-    k::Int                      = 3::(_ > 0)
-    tol::Float                  = 1e-6::(_ < 1)
-    max_iters::Int              = 300::(_ > 0)
+    algo::ParallelKMeans.AbstractKMeansAlg  = Lloyd()::(_ in (Lloyd(), Hamerly(), LightElkan()))
+    k_init::String                          = "k-means++"::(_ in ("k-means++", String)) # allow user seeding?
+    k::Int                                  = 3::(_ > 0)
+    tol::Float64                            = 1e-6::(_ < 1)
+    max_iters::Int                          = 300::(_ > 0)
+    #transpose_type::String                  = "permute"::(_ in ("permute", "transpose"))
+    threads::Int                            = Threads.nthreads()::(_ > 0)
+    verbosity::Int                          = 0::(_ in (0, 1))
+    init = nothing
 end
 
 
+# Expose all instances of user specified structs and package artifcats.
+const KMeansModel = Union{ParaKMeans}
+const ParallelKMeans_Desc = "Parallel & lightning fast implementation of all variants of the KMeans clustering algorithm in native Julia."
+
+
 # TODO 3: implementation of fit, predict, and fitted_params of the model
 ####
 #### FIT FUNCTION
 ####
+"""
+    TODO 3.1: Docs
+"""
+function MLJModelInterface.fit(m::ParaKMeans, verbosity::Int, X)
+    # fit the specified struct as a ParaKMeans model
+
+    # assumes user supplied table with columns as features
+    DMatrix = MLJModelInterface.matrix(X; transpose=true)
+    
+    # fit model and get results
+    if m.verbosity > 0
+		fitresult = ParallelKMeans.kmeans(m.algo, DMatrix, m.k;
+	                                      n_threads = m.threads, k_init=m.k_init,
+	                                      max_iters=m.max_iters, tol=m.tol, init=m.init,
+										  verbose=true)
+    else
+		fitresult = ParallelKMeans.kmeans(m.algo, DMatrix, m.k;
+	                                      n_threads = m.threads, k_init=m.k_init,
+	                                      max_iters=m.max_iters, tol=m.tol, init=m.init,
+										  verbose=false)
+    end
+
+    cache = nothing
+    report = NamedTuple{}()
 
-function MLJModelInterface.fit(m::ParaKMeans, verbosity::Int, X, y, w=nothing)
-    # body ...
     return (fitresult, cache, report)
 end
 
 
-function MLJModelInterface.fitted_params(model::ParaKMeans, fitresult)
+"""
+    TODO 3.2: Docs
+"""
+function MLJModelInterface.fitted_params(model::KMeansModel, fitresult)
     # extract what's relevant from `fitresult`
-    # ...
+    centres = fitresult.centres
+    converged = fitresult.converged
+    iters = fitresult.iterations
+    totalcost = fitresult.totalcost
     # then return as a NamedTuple
-    return (learned_param1 = ..., learned_param2 = ...)
+    return (centres = centres, totalcost = totalcost, iterations = iters, converged = converged)
 end
 
 
 ####
 #### PREDICT FUNCTION
 ####
+"""
+    TODO 3.3: Docs
+"""
 function MLJModelInterface.predict(m::ParaKMeans, fitresult, Xnew)
     # ...
 end
@@ -48,10 +88,8 @@ end
 #### METADATA
 ####
 
-# TODO 4: metadata for the package and for each of your models
-const PARAKMEANS_MODELS = Union{ParaKMeans}
-
-metadata_pkg.(PARAKMEANS_MODELS,
+# TODO 4: metadata for the package and for each of the model interfaces
+metadata_pkg.(KMeansModel,
     name = "ParallelKMeans",
     uuid = "42b8e9d4-006b-409a-8472-7f34b3fb58af", # see your Project.toml
     url  = "https://github.com/PyDataBlog/ParallelKMeans.jl",  # URL to your package repo
@@ -61,12 +99,12 @@ metadata_pkg.(PARAKMEANS_MODELS,
 )
 
 
-# Metadata for ParaKMeans model
+# Metadata for ParaKMeans model interface
 metadata_model(ParaKMeans,
     input   = MLJModelInterface.Table(MLJModelInterface.Continuous),  # what input data is supported?           # for a supervised model, what target?
     output  = MLJModelInterface.Table(MLJModelInterface.Count),  # for an unsupervised, what output?
     weights = false,                                             # does the model support sample weights?
-    descr   = "Parallel & lightning fast implementation of all variants of the KMeans clustering algorithm in native Julia.",
+    descr   = ParallelKMeans_Desc,
 	path	= "ParallelKMeans.src.mlj_interface.ParaKMeans"
 	#path    = "YourPackage.SubModuleContainingModelStructDefinition.YourModel1"
     )