Empact-hierclust 0.2.1 → 0.2.2

data/Empact-hierclust.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = %q{Empact-hierclust}
-  s.version = "0.2.1"
+  s.version = "0.2.2"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Brandt Kurowski", "Ben Woosley"]
-  s.date = %q{2010-11-01}
+  s.date = %q{2010-11-02}
   s.description = %q{performs hierarchical clustering on points in Euclidian space}
   s.email = %q{ben.woosley@gmail.com}
   s.extra_rdoc_files = [

data/History.txt

@@ -1,3 +1,9 @@
+== 0.2.2 2010-11-01
+
+* 1 minor enhancement
+  * Flatten incoming Point coordinates so that we can pass them as an array as well
+  * Make Clusterer resolution and separation arguments hash-based
+
 == 0.2.1 2010-11-01
 
 * 1 minor enhancement

data/VERSION

@@ -1 +1 @@
-0.2.1
+0.2.2

data/lib/hierclust/cluster.rb

@@ -13,10 +13,10 @@ module Hierclust
     end
     
     # Returns the average coordinates of all items in this Cluster.
-    def coordinates
+    def coordinates(nils = nil)
       return nil if size == 0
       @coordinates ||= begin
-        coords = self.points.map {|p| p.coordinates }
+        coords = self.points.map {|p| p.coordinates(nils) }
         coords = coords.shift.zip(*coords)
         coords.map {|points| points.inject(0.0) {|sum, p| sum + p } / points.size }
       end

data/lib/hierclust/clusterer.rb

@@ -16,18 +16,19 @@ module Hierclust
     # clustered, but will be put into clusters based strictly on coordinates.
     # The clusters generated by this "pre-clustering" will then be
     # hierarchically clustered as normal.
-    def initialize(data, separation = nil, resolution = nil)
-      @separation = separation
-      @resolution = resolution
+    def initialize(data, options = {})
+      @separation = options.delete(:separation)
+      @resolution = options.delete(:resolution)
+      @nils = options.delete(:nils)
       @data = precluster(data)
-      @distances = Distances.new(@data)
+      @distances = Distances.new(@data, @nils)
     end
 
     # Calculates and returns the set of clusters.
     def clusters
       return @data if @separation && @distances.separation > @separation
       while @data.length > 1
-        @distances = Distances.new(@data)
+        @distances = Distances.new(@data, @nils)
         return @data if @separation && @distances.separation > @separation
         @data = find_cluster
       end

data/lib/hierclust/distances.rb

@@ -4,7 +4,7 @@ module Hierclust
     attr_reader :nearest, :outliers, :separation
 
     # Create a new Distances for the given +items+
-    def initialize(items)
+    def initialize(items, nils = nil)
       @items = items
       @separation = 0
       @nearest = []
@@ -12,7 +12,7 @@ module Hierclust
       while !items.empty?
         origin = items.shift
         items.each do |other|
-          distance = origin.distance_to(other)
+          distance = origin.distance_to(other, nils)
           if @separation == 0 or distance < @separation
             @separation = distance
             @nearest = [origin, other]

data/lib/hierclust/point.rb

@@ -2,18 +2,25 @@ module Hierclust
   # A Point represents a single point in n-dimensional space.
   class Point
     # x-coordinate
-    attr_accessor :coordinates
     attr_accessor :data
     
     # Create a new Point with the given coordinates.
     def initialize(*coordinates)
       @data = coordinates.last.is_a?(Hash) ? coordinates.pop : {}
-      @coordinates = coordinates
+      @coordinates = coordinates.flatten
     end
     
+    def coordinates(nils = nil)
+      if nils
+        @coordinates.map {|c| c || nils }
+      else
+        @coordinates
+      end
+    end
+
     # Returns this distance from this Point to an +other+ Point.
-    def distance_to(other)
-      sum_of_squares = coordinates.zip(other.coordinates).map do |point, other_point|
+    def distance_to(other, nils = nil)
+      sum_of_squares = coordinates(nils).zip(other.coordinates(nils)).map do |point, other_point|
         (other_point - point) ** 2
       end.inject(0) {|sum, distance| sum + distance }
       Math.sqrt(sum_of_squares)

data/spec/hierclust/cluster_spec.rb

@@ -64,6 +64,70 @@ module Hierclust
     end
   end
 
+  describe Cluster, " with two points and data" do
+    before do
+      @x_1, @x_2 = 5, 15
+      @y_1, @y_2 = 4, 8
+      @p_1 = Point.new(@x_1, @y_1, :name => 'foo')
+      @p_2 = Point.new(@x_2, @y_2, :name => 'bar')
+      @c = Cluster.new([@p_1, @p_2])
+      @points = @c.points
+    end
+
+    it "should have coordinates at the average of points' coordinates" do
+      @c.coordinates.should == [10, 6]
+    end
+
+    it "should have two points" do
+      @points.size.should == 2
+    end
+
+    it "should include both points" do
+      @points.should include(@p_1, @p_2)
+    end
+
+    it "should retain the data on the points" do
+      @points.map(&:data).should =~ [{:name => 'foo'}, {:name => 'bar'}]
+    end
+
+    it "should have correct radius" do
+      radius = Math.sqrt((@x_1 - @x_2) ** 2 + (@y_1 - @y_2) ** 2) / 2.0
+      @c.radius.should == radius
+    end
+  end
+
+  describe Cluster, " with an array of two points and data" do
+    before do
+      @x_1, @x_2 = 5, 15
+      @y_1, @y_2 = 4, 8
+      @p_1 = Point.new([@x_1, @y_1], :name => 'foo')
+      @p_2 = Point.new([@x_2, @y_2], :name => 'bar')
+      @c = Cluster.new([@p_1, @p_2])
+      @points = @c.points
+    end
+
+    it "should have coordinates at the average of points' coordinates" do
+      @c.coordinates.should == [10, 6]
+    end
+
+    it "should have two points" do
+      @points.size.should == 2
+    end
+
+    it "should include both points" do
+      @points.should include(@p_1, @p_2)
+    end
+
+    it "should retain the data on the points" do
+      @points.map(&:data).should =~ [{:name => 'foo'}, {:name => 'bar'}]
+    end
+
+    it "should have correct radius" do
+      radius = Math.sqrt((@x_1 - @x_2) ** 2 + (@y_1 - @y_2) ** 2) / 2.0
+      @c.radius.should == radius
+    end
+  end
+
   describe Cluster, " with one point and one cluster" do
     before do
       @x_1, @x_2, @x_3 = 1, 2, 3

data/spec/hierclust/clusterer_spec.rb

@@ -110,7 +110,7 @@ module Hierclust
 
       describe "and separation 1" do
         before do
-          @c = Clusterer.new(@points, 1)
+          @c = Clusterer.new(@points, :separation => 1)
         end
 
         it "should return all four individual points" do
@@ -120,7 +120,7 @@ module Hierclust
 
       describe "and separation 2" do
         before do
-          @c = Clusterer.new(@points, 2)
+          @c = Clusterer.new(@points, :separation => 2)
         end
 
         it "should return two clusters" do
@@ -129,6 +129,35 @@ module Hierclust
       end
     end
 
+    describe "with missing values" do
+      before do
+        @points = [
+          Point.new(0, 1),
+          Point.new(1, 0),
+          Point.new(nil, 4),
+          Point.new(4, 3),
+        ]
+      end
+
+      describe "and no stand-in" do
+        it "should raise and error" do
+          lambda {
+            Clusterer.new(@points)
+          }.should raise_error
+        end
+      end
+
+      describe "with a stand-in" do
+        before do
+          @c = Clusterer.new(@points, :nils => 3, :separation => 1)
+        end
+        
+        it "should cluster as though the missing data were the stand-in" do
+          @c.clusters.size.should == 4
+        end
+      end
+    end
+
     describe "with eight points" do
       before do
         @points = [
@@ -155,7 +184,7 @@ module Hierclust
 
       describe "and separation 1" do
         before do
-          @clusters = Clusterer.new(@points, 1).clusters.sort
+          @clusters = Clusterer.new(@points, :separation => 1).clusters.sort
         end
 
         it "should have all eight points in individual clusters" do
@@ -166,7 +195,7 @@ module Hierclust
       describe "and separation 3" do
         describe "with no resolution limit" do
           before do
-            @clusters = Clusterer.new(@points, 3).clusters.sort
+            @clusters = Clusterer.new(@points, :separation => 3).clusters.sort
           end
 
           it "should have three clusters" do
@@ -186,7 +215,7 @@ module Hierclust
 
         describe "with coarse resolution" do
           before do
-            @clusters = Clusterer.new(@points, 3, 5).clusters.sort
+            @clusters = Clusterer.new(@points, :separation => 3, :resolution => 5).clusters.sort
           end
 
           it "should have three clusters" do

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: Empact-hierclust
 version: !ruby/object:Gem::Version 
-  hash: 21
+  hash: 19
   prerelease: false
   segments: 
   - 0
   - 2
-  - 1
-  version: 0.2.1
+  - 2
+  version: 0.2.2
 platform: ruby
 authors: 
 - Brandt Kurowski
@@ -16,7 +16,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2010-11-01 00:00:00 -07:00
+date: 2010-11-02 00:00:00 -07:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency