kmeans-clusterer 0.9.0 → 0.10.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9b1be676c0018dc27d0d3133b2ea021cf6349814
-  data.tar.gz: 0ef6ad8299f561accd85006569928b8b2d005a94
+  metadata.gz: 873f13d79f2d199400d1e359aca6e56c74d68ac0
+  data.tar.gz: a4d1f06a71b2e8289c60badeb93fb007fee06af6
 SHA512:
-  metadata.gz: aed54a15689e0b2c785a2d8aca57026959942239ca44576d3ffc835a84981a190daf1b0679580f15b8944a2f58b33c9f98ef60b59598781d65278e506c60ed45
-  data.tar.gz: 758a21a859d09e981c350d02de7965e39294f4dba667a741b08bab93fab8670a4d8573617c7d0896112c38f7a1e0be4723f9e20fb00a5f190bde55e89e8517e1
+  metadata.gz: 1e160f4bbe512e7aac37c2edcbcc63c1f103e51ae4c441a0c2f2e6cc57783344eddbcd31f30ea1a9a32dcc979250fb5bab6e1faba7a1d686612362d4aa798129
+  data.tar.gz: 405a6ab16edcb0fcdd958c9963446a44bd2954be9cdb1e95dc728bf7d10e18536334c693ca38f00b956d51358b3ab105aa2a9a3f68c862ab99914e4e4ee894e4

data/lib/kmeans-clusterer.rb

@@ -21,13 +21,18 @@ class KMeansClusterer
       data = (data - mean) / std
       [NMatrix.ref(data), mean, std]
     end
+
+    def self.row_norms data
+      squared_data = NArray.ref(data)**2
+      NMatrix.ref(squared_data).sum(0)
+    end
   end
 
   module Distance
     def self.euclidean x, y, yy = nil
       if x.is_a?(NMatrix) && y.is_a?(NMatrix)
-        xx = x.map {|v| v**2}.sum(0)
-        yy ||= y.map {|v| v**2}.sum(0)
+        xx = Scaler.row_norms(x)
+        yy ||= Scaler.row_norms(y)
         xy = x * y.transpose
         distance = xy * -2
         distance += xx
@@ -93,7 +98,7 @@ class KMeansClusterer
   end
 
 
-  DEFAULT_OPTS = { scale_data: false, runs: 10, log: false, init: :kmpp, float_precision: :double }
+  DEFAULT_OPTS = { scale_data: false, runs: 10, log: false, init: :kmpp, float_precision: :double, max_iter: 300 }
 
   def self.run k, data, opts = {}
     opts = DEFAULT_OPTS.merge(opts)
@@ -112,7 +117,7 @@ class KMeansClusterer
     end
 
     opts[:points_matrix] = data
-    opts[:row_norms] = opts[:points_matrix].map {|v| v**2}.sum(0)
+    opts[:row_norms] = Scaler.row_norms(data)
 
     bestrun = nil
 
@@ -147,6 +152,7 @@ class KMeansClusterer
     @std = opts[:std]
     @scale_data = opts[:scale_data]
     @typecode = opts[:typecode]
+    @max_iter = opts[:max_iter]
 
     init_centroids
   end
@@ -154,55 +160,60 @@ class KMeansClusterer
   def run 
     start_time = Time.now
     @iterations, @runtime = 0, 0
-
-    @cluster_point_ids = Array.new(@k) { [] }
+    @cluster_assigns = NArray.int(@points_count)
+    min_distances = NArray.new(@typecode, @points_count)
 
     loop do
       @iterations +=1
 
+      min_distances.fill! Float::INFINITY
       distances = Distance.euclidean(@centroids, @points_matrix, @row_norms)
 
-      # assign point ids to @cluster_point_ids
-      @points_count.times do |i|
-        min_distance_index = distances[i, true].sort_index[0]
-        @cluster_point_ids[min_distance_index] << i
+      @k.times do |cluster_id|
+        dist = NArray.ref distances[true, cluster_id].flatten
+        mask = dist < min_distances
+        @cluster_assigns[mask] = cluster_id
+        min_distances[mask] = dist[mask]
       end
 
-      moves = []
-      updated_centroids = []
+      max_move = 0
 
-      @k.times do |i|
-        centroid = NArray.ref(@centroids[true, i].flatten)
-        point_ids = @cluster_point_ids[i]
+      @k.times do |cluster_id|
+        centroid = NArray.ref(@centroids[true, cluster_id].flatten)
+        point_ids = @cluster_assigns.eq(cluster_id).where
 
-        if point_ids.empty?
-          newcenter = centroid
-          moves << 0
-        else
+        unless point_ids.empty?
           points = @points_matrix[true, point_ids]
           newcenter = points.mean(1)
-          moves << Distance.euclidean(centroid, newcenter)
+          move = Distance.euclidean(centroid, newcenter)
+          max_move = move if move > max_move
+          @centroids[true, cluster_id] = newcenter
         end
-
-        updated_centroids << newcenter
       end
 
-      @centroids = NMatrix.cast updated_centroids, @typecode
-
-      break if moves.max < 0.001 # i.e., no movement
-      break if @iterations >= 300
-
-      @cluster_point_ids = Array.new(@k) { [] }
+      break if max_move < 0.001 # i.e., no movement
+      break if @iterations >= @max_iter
     end
 
-    @error = calculate_error
+    @error = (min_distances**2).sum
     @runtime =  Time.now - start_time
     self
   end
 
   def finish
-    set_points
-    set_clusters
+    @clusters = @k.times.map do |i|
+      centroid = NArray.ref @centroids[true, i].flatten
+      Cluster.new i, Point.new(-i, centroid)
+    end
+
+    @points = @points_count.times.map do |i|
+      data = NArray.ref @points_matrix[true, i].flatten
+      point = Point.new(i, data, @labels[i])
+      cluster = @clusters[@cluster_assigns[i]]
+      cluster.points << point
+      point
+    end
+
     self
   end
 
@@ -305,40 +316,6 @@ class KMeansClusterer
       @points_count.times.to_a.sample @k
     end
 
-    def set_points
-      @points = @points_count.times.map do |i|
-        data = NArray.ref @points_matrix[true, i].flatten
-        Point.new(i, data, @labels[i])
-      end
-    end
-
-    def set_clusters
-      @clusters = @k.times.map do |i|
-        centroid = NArray.ref @centroids[true, i].flatten
-        c = Cluster.new i, Point.new(-i, centroid)
-        @cluster_point_ids[i].each do |p|
-          c << @points[p]
-        end
-        c
-      end
-    end
-
-    def calculate_error
-      errors = @k.times.map do |i|
-        centroid = get_centroid i
-        points = get_points_for_centroid i
-
-        if points.empty?
-          0
-        else
-          distances = Distance.euclidean points, centroid
-          (distances**2).sum
-        end
-      end
-
-      errors.reduce(:+)
-    end
-
     def get_point i
       NArray.ref @points_matrix[true, i].flatten
     end
@@ -348,7 +325,7 @@ class KMeansClusterer
     end
 
     def get_points_for_centroid i
-      point_ids = @cluster_point_ids[i]
+      point_ids = @cluster_assigns.eq(i).where
       points = @points_matrix[true, point_ids]
       points.empty? ? NArray.sfloat(0) : NArray.ref(points)
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: kmeans-clusterer
 version: !ruby/object:Gem::Version
-  version: 0.9.0
+  version: 0.10.0
 platform: ruby
 authors:
 - Geoff Buesing
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-03-03 00:00:00.000000000 Z
+date: 2015-03-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: narray