kmeans-clustering 1.0.0

data/lib/kmeans-clustering.rb

@@ -0,0 +1,118 @@
+module KMeansClustering
+  require 'cabiri'
+
+  # add static attributes through attr_accessor
+  class << self
+    attr_accessor :calcSum
+    attr_accessor :calcAverage
+    attr_accessor :calcDistanceSquared
+  end
+
+  # split array into several equal sized parts
+  # taken from http://apidock.com/rails/v3.2.8/Array/in_groups
+  def self.split_array_into_parts(array, nb_parts)
+    start = 0
+    groups = []
+
+    modulo = array.size % nb_parts
+    division = array.size / nb_parts
+
+    nb_parts.times do |index|
+      length = division + (modulo > 0 && modulo > index ? 1 : 0)
+      groups << array.slice(start, length)
+      start += length
+    end
+    groups
+  end
+
+  def self.run(centers, elements, nb_iterations, nb_jobs)
+    nb_iterations.times do
+      # create jobs
+      jobs = []
+      elements_for_jobs = split_array_into_parts(elements, nb_jobs)
+      nb_jobs.times do |i|
+        jobs << Job.new(centers, elements_for_jobs[i])
+      end
+
+      # run jobs in parallel
+      queue = Cabiri::JobQueue.new
+      nb_jobs.times do |i|
+        queue.add(i) { jobs[i].run }
+      end
+      queue.start(nb_jobs)
+
+      # sort aggregated proximity data by center
+      sorted_aggregated_proximity_data = Hash.new { |h,k| h[k] = [] }
+
+      queue.finished_jobs.values.each do |finished_job|
+        aggregated_proximity_data = finished_job.result
+        aggregated_proximity_data.each do |center, aggregated_data|
+          sorted_aggregated_proximity_data[center] << aggregated_data
+        end
+      end
+
+      # calculate sum and nb elements for each center
+      sums = Hash.new { |h,k| h[k] = [] }
+      nb_elements = Hash.new { |h,k| h[k] = [] }
+
+      sorted_aggregated_proximity_data.each do |center, aggregated_data|
+        sums[center] = KMeansClustering::calcSum.call(aggregated_data.collect { |d| d[:sum] })
+        nb_elements[center] = (aggregated_data.collect { |d| d[:nb_elements] }).inject(0, :+)
+      end
+
+      # calculate new centers
+      centers = []
+      sums.keys.each do |center|
+        centers << KMeansClustering::calcAverage.call(sums[center], nb_elements[center])
+      end
+    end
+
+    centers
+  end
+
+  # job that will be used for parallelization with Cabiri
+  class Job
+    attr_accessor :centers
+    attr_accessor :elements
+
+    def initialize(centers, elements)
+      @centers = centers
+      @elements = elements
+    end
+
+    def run
+      proximity_data = assignElementsToClosestCenter
+      aggregated_proximity_data = aggregateProximityData(proximity_data)
+      aggregated_proximity_data
+    end
+
+    def assignElementsToClosestCenter
+      results = Hash.new { |h,k| h[k] = [] }
+
+      @elements.each do |element|
+        best_center = nil
+        best_distance = nil
+
+        @centers.each do |center|
+          distance = KMeansClustering::calcDistanceSquared.call(center, element)
+          if best_distance.nil? or distance < best_distance
+            best_center = center
+            best_distance = distance
+          end
+        end
+        results[best_center] << element
+      end
+
+      results
+    end
+
+    def aggregateProximityData(data)
+      results = {}
+      data.each do |center, elements|
+        sum = KMeansClustering::calcSum.call(elements)
+        results[center] = {:sum => sum, :nb_elements => elements.length}
+      end
+      results
+    end
+  end
+end

metadata

@@ -0,0 +1,62 @@
+--- !ruby/object:Gem::Specification
+name: kmeans-clustering
+version: !ruby/object:Gem::Version
+  version: 1.0.0
+  prerelease: 
+platform: ruby
+authors:
+- Tom Van Eyck
+autorequire: 
+bindir: bin
+cert_chain: []
+date: 2013-02-17 00:00:00.000000000 Z
+dependencies:
+- !ruby/object:Gem::Dependency
+  name: cabiri
+  requirement: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: 0.0.7
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ! '>='
+      - !ruby/object:Gem::Version
+        version: 0.0.7
+description: A simple Ruby gem for parallelized k-means clustering.
+email: tomvaneyck@gmail.com
+executables: []
+extensions: []
+extra_rdoc_files: []
+files:
+- lib/kmeans-clustering.rb
+homepage: https://github.com/vaneyckt/kmeans-clustering
+licenses: []
+post_install_message: 
+rdoc_options: []
+require_paths:
+- lib
+required_ruby_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+required_rubygems_version: !ruby/object:Gem::Requirement
+  none: false
+  requirements:
+  - - ! '>='
+    - !ruby/object:Gem::Version
+      version: '0'
+requirements: []
+rubyforge_project: 
+rubygems_version: 1.8.24
+signing_key: 
+specification_version: 3
+summary: A simple Ruby gem for parallelized k-means clustering.
+test_files: []
+has_rdoc: