agglomerative_clustering 0.0.1 → 0.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: d52ff4f84aef0fbd56aa453833ea2e5b9c9772e2
-  data.tar.gz: 4abb19bb148f77989f43be805769e0676a7182f7
+  metadata.gz: 6368eb116e76afdf46d5dd16ffeaed04a3d590e8
+  data.tar.gz: 39447a07d51c37d5743dc910bb71ff011bfcd6d2
 SHA512:
-  metadata.gz: b31e61ad6c08ecdce2326bdab94c412aaee32227775f662ad62361896d69124acbad3ed4c6f8a40f2298de30ac77d65c0fcabd0bfe58c16b6d11bd89e75a7dd5
-  data.tar.gz: 8a2a08bda0f2975166536d1ec85cd0f2254044689d8b6f422b6f2170bd64301ae914c05f1264d14a81d7b4b2fb4deb9c4da0fa82abb9b94f57a462ba1689c11d
+  metadata.gz: 198911398132736c1cf06117e48e4f0026c94271586e8f172bf72733828f8bf1b218e03659b148663c56b2f8a3ca793e45b335527e90bc51577663221ddb4202
+  data.tar.gz: 5b04529399622524c8d7b36cd8b387c7dae73c83700e0e63455091e6357b39745bb78df0d01e6df0f8e29f0d264f2df89b23a79d9b744f2df393ceb10062249a

data/Gemfile

@@ -1,3 +1,5 @@
 source "https://rubygems.org"
 gemspec
-gem 'factory_girl'
+group :development do
+  gem 'factory_girl'
+end

data/README.md

@@ -2,7 +2,12 @@
 
 Hierarchical Agglomerative Clustering Algorithm
 
-Input Set of 3 dimensional points, group into nearest k clusters
+Input Set of 3 dimensional points, group into nearest k clusters based on Euclidean Distance.
+Currently the Clustering Algorithm supports 4 different types of Linkage
+* Single Linkage (Distance between clusters is based on nearest points)
+* Complete Linkage (Distance between clusters is based on farthest points)
+* Average Linkage (Distance between clusters is based on average distance of points)
+* Center Linkage (Distance between clusters is based on center of cluster)
 
 ## Installation
 
@@ -21,8 +26,8 @@ Or install it yourself as:
     $ gem install agglomerative_clustering
 
 ## Usage
+Please see cluster.rb for a sample until I have a chance to write something up here
 
-TODO: Write usage instructions here
 
 ## Contributing
 

data/agglomerative_clustering.gemspec

@@ -7,9 +7,9 @@ Gem::Specification.new do |spec|
   spec.name          = "agglomerative_clustering"
   spec.version       = AgglomerativeClustering::VERSION
   spec.authors       = ["Bryan Mulvihill"]
-  spec.email         = ["bmulvihill@pinsonault.com"]
+  spec.email         = ["mulvihill.bryan@gmail.com"]
   spec.summary       = %q{Ruby Agglomerative Clustering Algorithm}
-  spec.homepage      = ""
+  spec.homepage      = "https://github.com/bmulvihill/agglomerative_clustering"
   spec.license       = "MIT"
 
   spec.files         = `git ls-files -z`.split("\x0")

data/lib/agglomerative_clustering.rb

@@ -1,5 +1,6 @@
 require "agglomerative_clustering/version"
 require "agglomerative_clustering/euclidean_distance"
+require "agglomerative_clustering/distance_matrix"
 require "agglomerative_clustering/linkage/base"
 require "agglomerative_clustering/linkage/single"
 require "agglomerative_clustering/linkage/complete"

data/lib/agglomerative_clustering/distance_matrix.rb

@@ -0,0 +1,37 @@
+module AgglomerativeClustering
+  class DistanceMatrix
+
+    def initialize matrix
+      @matrix_array = matrix.to_a
+    end
+
+    def matrix
+      Matrix.build(matrix_array.size, matrix_array.first.size) do |row, column|
+        matrix_array[row][column]
+      end
+    end
+
+    def print_matrix
+      puts matrix.to_a.map(&:inspect)
+    end
+
+    def remove_edge index
+      matrix_array.delete_at(index)
+      matrix_array.each { |row| row.delete_at(index) }
+      Matrix.rows(matrix_array)
+    end
+
+    def add_edge weights
+      matrix_array.each_with_index { |row, index|  row << weights[index] }
+      matrix_array << weights
+      Matrix.rows(matrix_array)
+    end
+    
+    private
+
+    def matrix_array
+      @matrix_array ||= []
+    end
+
+  end
+end

data/lib/agglomerative_clustering/linkage/average.rb

@@ -12,9 +12,6 @@ module AgglomerativeClustering
         distances.inject(:+)/distances.size
       end
 
-      def clusters_to_merge
-        @clusters_to_merge ||= []
-      end
     end
   end
 end

data/lib/agglomerative_clustering/linkage/base.rb

@@ -2,21 +2,6 @@ module AgglomerativeClustering
   module Linkage
     class Base
       include EuclideanDistance
-
-      def cluster(clusters)
-        min_cluster_dist = 1.0/0
-        clusters.each_with_index do |cluster1, index|
-          clusters[index + 1..clusters.size].each do |cluster2|
-            distance = calculate_distance(cluster1, cluster2)
-            if distance < min_cluster_dist
-              min_cluster_dist = distance
-              @clusters_to_merge = [cluster1, cluster2]
-            end
-          end
-        end
-        clusters_to_merge
-      end
-      
     end
   end
 end

data/lib/agglomerative_clustering/linkage/center.rb

@@ -10,11 +10,7 @@ module AgglomerativeClustering
       def center_point cluster
         cluster.points.first.zip(*cluster.points[1..cluster.points.size-1]).map { |a,b| (a + b)/cluster.points.size.to_f }
       end
-
-      def clusters_to_merge
-        @clusters_to_merge ||= []
-      end
-
+      
     end
   end
 end

data/lib/agglomerative_clustering/linkage/complete.rb

@@ -13,10 +13,6 @@ module AgglomerativeClustering
         max_point_distance
       end
 
-      def clusters_to_merge
-        @clusters_to_merge ||= []
-      end
-
     end
   end
 end

data/lib/agglomerative_clustering/linkage/single.rb

@@ -13,9 +13,6 @@ module AgglomerativeClustering
         min_point_distance
       end
 
-      def clusters_to_merge
-        @clusters_to_merge ||= []
-      end
     end
   end
 end

data/lib/agglomerative_clustering/point.rb

@@ -1,3 +1 @@
-Point = Struct.new(:x, :y, :z) do
-  attr_accessor :index
-end
+Point = Struct.new(:x, :y, :z)

data/lib/agglomerative_clustering/set.rb

@@ -2,15 +2,17 @@ require 'matrix'
 module AgglomerativeClustering
   class Set
     include EuclideanDistance
-    attr_reader :points
+    attr_reader :linkage
 
     def initialize(linkage)
       @linkage = linkage
-      @points = []
+    end
+
+    def points
+      @points ||= []
     end
 
     def push point
-      point.index = points.size
       points << point
     end
 
@@ -22,23 +24,28 @@ module AgglomerativeClustering
       @distance_matrix ||= build_distance_matrix
     end
 
-    def print_distance_matrix
-      puts distance_matrix.to_a.map(&:inspect)
-    end
-
     def cluster total_clusters
-      clusters_to_merge =[]
       while clusters.size > total_clusters
-        clusters_to_merge = @linkage.cluster(clusters)
-        merge_clusters(clusters_to_merge)
+        merge_clusters(shortest_distance)
       end
       clusters
     end
 
-    def merge_clusters(min_clusters)
-      min_clusters[0].merge(min_clusters[1])
-      clusters.reject! { |cluster| cluster == min_clusters[1] }
-      min_clusters[0]
+    def merge_clusters indexes
+      index1, index2 = indexes
+      new_cluster = clusters[index1].merge(clusters[index2])
+      remove_cluster(index1)
+      remove_cluster(index2 - 1)
+      add_cluster(new_cluster)
+    end
+
+    def update_distance_matrix new_cluster
+      distances = []
+      clusters.each do |cluster|
+        distances << linkage.calculate_distance(clusters[new_cluster], cluster)
+      end
+      distance_matrix.add_edge(distances)
+      distance_matrix
     end
 
     def outliers
@@ -46,10 +53,10 @@ module AgglomerativeClustering
     end
 
     def find_outliers percentage_of_clusters, distance
-      distance_matrix.each_with_index do |index, row, column|
+      distance_matrix.matrix.each_with_index do |index, row, column|
         count_hash[row] ||= 0
-        count_hash[row] += 1 if distance_matrix[row, column] > distance
-        set_outliers << points[row] if count_hash[row]/(distance_matrix.row_count - 1) > percentage_of_clusters/100
+        count_hash[row] += 1 if distance_matrix.matrix[row, column] > distance
+        set_outliers << points[row] if count_hash[row]/(distance_matrix.matrix.row_count - 1) > percentage_of_clusters/100
       end
       points.reject! { |point| outliers.include?(point) }
       outliers
@@ -57,6 +64,30 @@ module AgglomerativeClustering
 
     private
 
+    def add_cluster new_cluster
+      clusters << new_cluster
+      update_distance_matrix(clusters.size - 1)
+      new_cluster
+    end
+
+    def remove_cluster index
+      clusters.delete_at(index)
+      distance_matrix.remove_edge(index)
+    end
+
+    def shortest_distance
+      min_cluster_dist = 1.0/0
+      indexes = []
+      distance_matrix.matrix.each_with_index do |index, row, column|
+        distance = distance_matrix.matrix[row, column]
+        if distance < min_cluster_dist && distance != 0
+          min_cluster_dist = distance
+          indexes = [row, column]
+        end
+      end
+      indexes
+    end
+
     def set_outliers
       @set_outliers ||= []
     end
@@ -66,9 +97,10 @@ module AgglomerativeClustering
     end
 
     def build_distance_matrix
-      Matrix.build(points.size, points.size) do |row, column|
+      m = Matrix.build(points.size, points.size) do |row, column|
         euclidean_distance(points[row], points[column]).round(2)
       end
+      DistanceMatrix.new(m)
     end
 
   end

data/lib/agglomerative_clustering/version.rb

@@ -1,3 +1,3 @@
 module AgglomerativeClustering
-  VERSION = "0.0.1"
+  VERSION = "0.0.2"
 end

data/spec/lib/agglomerative_clustering/distance_matrix_spec.rb

@@ -0,0 +1,18 @@
+describe AgglomerativeClustering::DistanceMatrix do
+
+  context '#remove_edge' do
+    it 'will remove edges from the distance matrix' do
+      matrix = AgglomerativeClustering::DistanceMatrix.new(Matrix.empty)
+      matrix.add_edge([1,2])
+      matrix.add_edge([2,2,3])
+      expect(matrix.remove_edge(0)).to eql(Matrix[[2,3]])
+    end
+  end
+
+  context '#add_edge' do
+    it 'will add an edges to the distance matrix' do
+      matrix = AgglomerativeClustering::DistanceMatrix.new(Matrix.rows([[1,2,3]]))
+      expect(matrix.add_edge([4,5,6,7])).to eql(Matrix[[1,2,3,4],[4,5,6,7]])
+    end
+  end
+end

data/spec/lib/agglomerative_clustering/set_spec.rb

@@ -2,9 +2,9 @@ describe AgglomerativeClustering::Set do
 
   before do
     @set = FactoryGirl.build(:set)
-    @point1 = FactoryGirl.build(:point, x:2, y:2, z:3)
-    @point2 = FactoryGirl.build(:point, x:1, y:4, z:1)
-    @point3 = FactoryGirl.build(:point, x:5, y:2, z:2)
+    @point1 = FactoryGirl.build(:point, x:1, y:2, z:3)
+    @point2 = FactoryGirl.build(:point, x:2, y:4, z:1)
+    @point3 = FactoryGirl.build(:point, x:4, y:2, z:2)
     @point4 = FactoryGirl.build(:point, x:5, y:2, z:3)
     @set.push(@point1)
     @set.push(@point2)
@@ -14,36 +14,30 @@ describe AgglomerativeClustering::Set do
 
   context '#cluster' do
     it 'will return clusters of points based on requested number of clusters' do
-      expect(@set.cluster(3).size).to eql(3)
+      expect(@set.cluster(2).size).to eql(2)
     end
 
     it 'will cluster points that are closest to each other' do
-      @point5 = FactoryGirl.build(:point, x:5, y:2, z:4)
-      @point6 = FactoryGirl.build(:point, x:5, y:3, z:4)
+      @point5 = FactoryGirl.build(:point, x:6, y:2, z:4)
+      @point6 = FactoryGirl.build(:point, x:7, y:3, z:4)
       @point7 = FactoryGirl.build(:point, x:15, y:20, z:21)
-      @point8 = FactoryGirl.build(:point, x:18, y:21, z:21)
-      @point9 = FactoryGirl.build(:point, x:16, y:22, z:21)
+      @point8 = FactoryGirl.build(:point, x:16, y:21, z:21)
+      @point9 = FactoryGirl.build(:point, x:18, y:22, z:21)
       @set.push(@point5)
       @set.push(@point6)
       @set.push(@point7)
       @set.push(@point8)
       @set.push(@point9)
       clusters = @set.cluster(3)
-      clusters[0].points.each do |point|
-        expect([@point1, @point2].include?(point)).to be true
-      end
-      clusters[1].points.each do |point|
-        expect([@point3, @point4, @point5, @point6].include?(point)).to be true
-      end
-      clusters[2].points.each do |point|
-        expect([@point7, @point8, @point9].include?(point)).to be true
-      end
+      points = clusters.map(&:points).each {|cluster| cluster.sort_by!(&:x) }
+      expect([[@point1, @point2],[@point3, @point4, @point5, @point6], [@point7, @point8, @point9]] - points).to eql([])
+
     end
   end
 
   context '#merge_clusters' do
     it 'will merge two clusters into one and update the distance matrix' do
-      expect(@set.merge_clusters([@set.clusters[0],@set.clusters[1]]).points).to eql([@point1, @point2])
+      expect(@set.merge_clusters([0,1]).points).to eql([@point1, @point2])
     end
   end
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: agglomerative_clustering
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors:
 - Bryan Mulvihill
@@ -54,7 +54,7 @@ dependencies:
         version: '0'
 description: 
 email:
-- bmulvihill@pinsonault.com
+- mulvihill.bryan@gmail.com
 executables: []
 extensions: []
 extra_rdoc_files: []
@@ -63,7 +63,6 @@ files:
 - ".rspec"
 - Gemfile
 - Gemfile.lock
-- LICENSE
 - LICENSE.txt
 - README.md
 - Rakefile
@@ -71,6 +70,7 @@ files:
 - cluster.rb
 - lib/agglomerative_clustering.rb
 - lib/agglomerative_clustering/cluster.rb
+- lib/agglomerative_clustering/distance_matrix.rb
 - lib/agglomerative_clustering/euclidean_distance.rb
 - lib/agglomerative_clustering/linkage/average.rb
 - lib/agglomerative_clustering/linkage/base.rb
@@ -80,21 +80,19 @@ files:
 - lib/agglomerative_clustering/point.rb
 - lib/agglomerative_clustering/set.rb
 - lib/agglomerative_clustering/version.rb
-- outliers.csv
-- points.csv
 - spec/factories/lib/agglomerative_clustering/cluster.rb
 - spec/factories/lib/agglomerative_clustering/point.rb
 - spec/factories/lib/agglomerative_clustering/set.rb
 - spec/lib/agglomerative_clustering/cluster_spec.rb
+- spec/lib/agglomerative_clustering/distance_matrix_spec.rb
 - spec/lib/agglomerative_clustering/euclidean_distance_spec.rb
 - spec/lib/agglomerative_clustering/linkage/average_spec.rb
-- spec/lib/agglomerative_clustering/linkage/base_spec.rb
 - spec/lib/agglomerative_clustering/linkage/center_spec.rb
 - spec/lib/agglomerative_clustering/linkage/complete_spec.rb
 - spec/lib/agglomerative_clustering/linkage/single_spec.rb
 - spec/lib/agglomerative_clustering/set_spec.rb
 - spec/spec_helper.rb
-homepage: ''
+homepage: https://github.com/bmulvihill/agglomerative_clustering
 licenses:
 - MIT
 metadata: {}
@@ -123,9 +121,9 @@ test_files:
 - spec/factories/lib/agglomerative_clustering/point.rb
 - spec/factories/lib/agglomerative_clustering/set.rb
 - spec/lib/agglomerative_clustering/cluster_spec.rb
+- spec/lib/agglomerative_clustering/distance_matrix_spec.rb
 - spec/lib/agglomerative_clustering/euclidean_distance_spec.rb
 - spec/lib/agglomerative_clustering/linkage/average_spec.rb
-- spec/lib/agglomerative_clustering/linkage/base_spec.rb
 - spec/lib/agglomerative_clustering/linkage/center_spec.rb
 - spec/lib/agglomerative_clustering/linkage/complete_spec.rb
 - spec/lib/agglomerative_clustering/linkage/single_spec.rb

data/LICENSE

@@ -1,22 +0,0 @@
-The MIT License (MIT)
-
-Copyright (c) 2014 Bryan Mulvihill
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
-

data/spec/lib/agglomerative_clustering/linkage/base_spec.rb

@@ -1,13 +0,0 @@
-describe AgglomerativeClustering::Linkage::Base do
-
-  context '#cluster' do
-    it 'will return the clusters where min distance is closest' do
-      single_linkage = AgglomerativeClustering::Linkage::Single.new
-      set = FactoryGirl.build(:set)
-      set.push(FactoryGirl.build(:point))
-      set.push(FactoryGirl.build(:point))
-      expect(single_linkage.cluster(set.clusters)).to eql([set.clusters[0], set.clusters[1]])
-    end
-  end
-
-end