Merge pull request #23 from JuliaAI/adjoint-not-transpose

Use adjoints not transposes

Author: ablaom

Project.toml

@@ -1,7 +1,7 @@
 name = "MLJClusteringInterface"
 uuid = "d354fa79-ed1c-40d4-88ef-b8c7bd1568af"
 authors = ["Anthony D. Blaom <[email protected]>", "Thibaut Lienart <[email protected]>", "Okon Samuel <[email protected]>"]
-version = "0.1.8"
+version = "0.1.9"
 
 [deps]
 Clustering = "aaaa29a8-35af-508c-8bc3-b662a17a0fe5"

src/MLJClusteringInterface.jl

@@ -35,9 +35,7 @@ const PKG = "MLJClusteringInterface"
 end
 
 function MMI.fit(model::KMeans, verbosity::Int, X)
-    # NOTE: using transpose here to get a LinearAlgebra.Transpose object
-    # which Kmeans can handle.
-    Xarray = transpose(MMI.matrix(X))
+    Xarray = MMI.matrix(X)'
     result = Cl.kmeans(Xarray, model.k; distance=model.metric, init=model.init)
     cluster_labels = MMI.categorical(1:model.k)
     fitresult = (result.centers, cluster_labels) # centers (p x k)
@@ -55,7 +53,7 @@ function MMI.transform(model::KMeans, fitresult, X)
     # pairwise distance from samples to centers
     X̃ = pairwise(
         model.metric,
-        transpose(MMI.matrix(X)),
+        MMI.matrix(X)',
         fitresult[1],
         dims=2
     )
@@ -71,9 +69,7 @@ end
 end
 
 function MMI.fit(model::KMedoids, verbosity::Int, X)
-    # NOTE: using transpose=true will materialize the transpose (~ permutedims), KMedoids
-    # does not yet accept LinearAlgebra.Transpose
-    Xarray = MMI.matrix(X, transpose=true)
+    Xarray = MMI.matrix(X)'
     # cost matrix: all the pairwise distances
     cost_array = pairwise(model.metric, Xarray, dims=2) # n x n
     result = Cl.kmedoids(cost_array, model.k, init = model.init)
@@ -93,7 +89,7 @@ function MMI.transform(model::KMedoids, fitresult, X)
     # pairwise distance from samples to medoids
     X̃ = pairwise(
         model.metric,
-        MMI.matrix(X, transpose=true),
+        MMI.matrix(X)',
         fitresult[1], dims=2
     )
     return MMI.table(X̃, prototype=X)

test/Project.toml

@@ -3,9 +3,6 @@ Clustering = "aaaa29a8-35af-508c-8bc3-b662a17a0fe5"
 Distances = "b4f34e82-e78d-54a5-968a-f98e89d6e8f7"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 MLJBase = "a7f614a8-145f-11e9-1d2a-a57a1082229d"
-MLJTestIntegration = "697918b4-fdc1-4f9e-8ff9-929724cee270"
+MLJTestInterface = "72560011-54dd-4dc2-94f3-c5de45b75ecd"
 Random = "9a3f8284-a2c9-5f02-9a11-845980a1fd5c"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
-
-[compat]
-MLJTestIntegration = "0.2.2"

test/runtests.jl

@@ -3,7 +3,7 @@ import Distances
 import LinearAlgebra: norm
 
 using MLJBase
-using MLJTestIntegration
+using MLJTestInterface
 using MLJClusteringInterface
 using Random: seed!
 using Test
@@ -122,11 +122,11 @@ end
 
 @testset "MLJ interface" begin
     models = [KMeans, KMedoids, DBSCAN, HierarchicalClustering]
-    failures, summary = MLJTestIntegration.test(
+    failures, summary = MLJTestInterface.test(
         models,
         X;
         mod=@__MODULE__,
-        verbosity=0,
+        verbosity=0, # bump to debug
         throw=false, # set to true to debug
     )
     @test isempty(failures)