kmeans-crystal 1.0.1 → 1.1.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 92da21fd5b192c00cb575b352410fc09a0cf3649
-  data.tar.gz: 72563e5d97d4a76e58fd128b83894510632e2fea
+  metadata.gz: c25646b44067f4a43cca87296ef5d04ff77272c5
+  data.tar.gz: 83e65a93b07dff510d80df973affc7671f565198
 SHA512:
-  metadata.gz: 6b2e361cfd58f4d6248fd6f6d1136058b555939a2ea5ca617980e809bcab60eeb3ea77a48f662ddfea4edfbb8e7e546edc541de5a66784b5d2f35a2788516fb1
-  data.tar.gz: 77a92b7b706f01641497976f0e9c0a092a800f358cd01cf13ce9b9ebcd3d331529d6bd45df43da2fa4caca5246adf2cf051bd07e0d357c6aecfd273eacbb5980
+  metadata.gz: 52d5da1622b0362238fab40aa34b64a948e381066afc90547f52747aed3220d21cb0980879790bc16bc0a0e9e685b33cef880067bd879c3fae88f17010928837
+  data.tar.gz: 719f3197099ad25d359d1758a48cfcd3fe7913f332d31fc931a291075ae795b555dfaa35b2968793dcb170bedb1d34f0ebc9af28dacb8aee05b79c97f2bf8e3c

data/lib/kmeans-crystal.rb

@@ -1,24 +1,46 @@
 module KMeansCrystal
+
+module Measure
+    # 歐式距離
+    class Euclidean
+        def self.distance(a,b)
+            sum = 0.0
+            a.size.times{|i| sum += (a[i] -b[i])**2 }
+            return Math.sqrt(sum)
+        end
+    end
+
+    # 曼哈頓距離
+    class Manhattan
+        def self.distance(a,b)
+            sum = 0.0
+            a.size.times{|i| sum += (a[i] -b[i]).abs }
+            return Math.sqrt(sum)
+        end
+    end
+end
+
+
 class Cluster
     attr_reader :centroid
     attr_reader :entries
     attr_accessor :name
 
-    def initialize(name, centroid, vector_name)
+    def initialize(name, centroid, vector_name, measure)
         @name = name
         @centroid = centroid
         @entries = Array.new
         @vector_name = vector_name
+        @measure = measure
     end
 
     def output
-        return { name: @name, centroid: @centroid, entries: @entries }
+        output_entries = @entries.map{|e| e[:distance] = distance(e); e }
+        return { name: @name, centroid: @centroid, entries: output_entries }
     end
 
     def distance(entry)
-        sum = 0.0
-        @centroid.size.times{|i| sum += (@centroid[i]-entry[@vector_name][i])**2}
-        return Math.sqrt(sum)
+        return @measure.distance(@centroid, entry[@vector_name])
     end
 
     def update_centroid
@@ -37,13 +59,37 @@ end
 
 
 class Model
-    def initialize(cluster_num, entries, vector_name = :features)
+    def initialize(cluster_num, entries, **params)
         raise 'too less cluster_num to evaluate k-means' if entries.size < cluster_num
+
         @cluster_num = cluster_num
         @entries = entries
-        @vector_name = vector_name
 
-        init_centroids = @entries.sample(@cluster_num).map{|x| x[@vector_name]}
+        @vector_name = case params[:vector_name]
+        when nil
+            :features
+        else
+            params[:vector_name]
+        end
+
+        @measure = case params[:distance]
+        when 'manhattan'
+            Measure::Manhattan
+        when 'euclidean',nil
+            Measure::Euclidean
+        else
+            raise 'incorrect value for distance'
+        end
+
+        init_centroids = case params[:init_centroids]
+        when 'random'
+            @entries.sample(@cluster_num).map{|x| x[@vector_name]}
+        when 'kmeans++',nil
+            kmeans_pp(@entries, @cluster_num)
+        else
+            raise 'incorrect value for init_centroids'
+        end
+
         @clusters = new_clusters(init_centroids)
     end
 
@@ -96,7 +142,7 @@ class Model
     def new_clusters(centroids)
         clusters = Array.new
         centroids.each_with_index do |centroid, i|
-            clusters << Cluster.new("cluster#{i}", centroid, @vector_name)
+            clusters << Cluster.new("cluster#{i}", centroid, @vector_name, @measure)
         end
         return clusters
     end
@@ -104,9 +150,32 @@ class Model
     def new_clusters_from_old(clusters)
         arr = Array.new
         clusters.each do |cluster|
-            arr << Cluster.new(cluster.name, cluster.update_centroid, @vector_name)
+            arr << Cluster.new(cluster.name, cluster.update_centroid, @vector_name, @measure)
         end
         return arr
     end
+
+    def kmeans_pp(entries, cluster_num)
+        features = entries.map{|x| x[@vector_name]}
+        dimension = features.first.size
+        init_val = Array.new(dimension){[0.0,0.0]}
+        dimension.times do |i|
+            init_val[i][0] = features.map{|f| f[i]}.to_a.min
+            init_val[i][1] = features.map{|f| f[i]}.to_a.max
+        end
+
+        init_centroids = entries.sample(cluster_num).map{|x| x[@vector_name]}
+        combination = 2**dimension
+        combination.times do |i|
+            break if i >= init_centroids.size
+            offset = i.to_s(2).rjust(dimension,"0")
+            dimension.times do |d|
+                max_or_min = offset[d].to_i
+                init_centroids[i][d] = init_val[d][max_or_min]
+            end
+
+        end
+        return init_centroids
+    end
 end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: kmeans-crystal
 version: !ruby/object:Gem::Version
-  version: 1.0.1
+  version: 1.1.0
 platform: ruby
 authors:
 - ireullin
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2016-08-05 00:00:00.000000000 Z
+date: 2017-04-14 00:00:00.000000000 Z
 dependencies: []
 description: |-
   The library for data clustering is implemented by k-means algorithm.With the library, you can monitor the model’s training processand end the training if the result is converged.
@@ -40,10 +40,9 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.4.6
+rubygems_version: 2.2.2
 signing_key: 
 specification_version: 4
 summary: With this library, you can monitor the model’s training process and end the
   training if the result is converged.
 test_files: []
-has_rdoc: