hierclust 0.1.1 → 0.1.2

data/History.txt

@@ -1,3 +1,10 @@
+== 0.1.2 2008-02-07
+
+* 1 performance improvement
+  * refactored Distances to be more intelligent about precalculated values
+* 1 bugfix:
+  * correct degenerate case of Hierclust::Point.points
+
 == 0.1.1 2008-02-04
 
 * 1 minor enhancement:

data/lib/hierclust/distances.rb

@@ -1,56 +1,25 @@
 module Hierclust
   # Represents the pair-wise distances between a set of items.
   class Distances
+    attr_reader :nearest, :outliers, :separation
+
     # Create a new Distances for the given +items+
     def initialize(items)
       @items = items
-      @distances = {}
-    end
-
-    # Returns the distance between items +a+ and +b+.
-    def [](a, b)
-      key = (a.object_id < b.object_id) ?
-            "#{a.object_id},#{b.object_id}" :
-            "#{b.object_id},#{a.object_id}"
-      @distances[key] ||= Math.sqrt((a.x - b.x) ** 2 + (a.y - b.y) ** 2)  
-    end
-
-    # Returns the pair of items that are nearest to each other.
-    def nearest
-      @nearest ||= begin
-        key = nearest_pair[0]
-        a_id, b_id = key.split ','
-        a = @items.select {|item| item.object_id.to_s == a_id}.first
-        b = @items.select {|item| item.object_id.to_s == b_id}.first
-        [a, b]
-      end
-    end
-    
-    # Returns all items except the pair that are nearest to each other.
-    def outliers
-      @outliers ||= begin
-        @items - nearest
-      end
-    end
-    
-    # Returns the smallest distance between any pair of items.
-    def separation
-      @separation ||= @items.size < 2 ? 0 : nearest_pair[1]
-    end
-
-    private
-
-    def nearest_pair
-      prepopulate_cache
-      @distances.to_a.sort{|a, b| a[1] <=> b[1]}.first
-    end
-    
-    def prepopulate_cache
+      @separation = 0
+      @nearest = []
       items = @items.dup
       while !items.empty?
         origin = items.shift
-        items.each {|item| self[origin, item]}
+        items.each do |other|
+          distance = origin.distance_to(other)
+          if @separation == 0 or distance < @separation
+            @separation = distance
+            @nearest = [origin, other]
+          end
+        end
       end
+      @outliers = @items - @nearest
     end
   end
 end

data/lib/hierclust/point.rb

@@ -12,7 +12,12 @@ module Hierclust
       @x = x
       @y = y
     end
-
+    
+    # Returns this distance from this Point to an +other+ Point.
+    def distance_to(other)
+      Math.sqrt((other.x - self.x) ** 2 + (other.y - self.y) ** 2)  
+    end
+    
     # Simplifies code by letting us treat Clusters and Points interchangeably
     def size #:nodoc:
       1
@@ -20,7 +25,7 @@ module Hierclust
     
     # Simplifies code by letting us treat Clusters and Points interchangeably
     def points #:nodoc:
-      self
+      [self]
     end
     
     # Returns a legible representation of this Point.
@@ -29,8 +34,17 @@ module Hierclust
     end
     
     # Sorts points relative to each other on the x-axis.
+    # 
+    # Uses y-axis as a tie-breaker, so that sorting is stable even if
+    # multiple points have the same x-coordinate.
+    # 
+    # Uses object_id as a final tie-breaker, so sorts are guaranteed to
+    # be stable even when multiple points have the same coordinates.
     def <=>(other)
-      return self.x <=> other.x
+      cmp = self.x <=> other.x
+      cmp = self.y <=> other.y if cmp == 0
+      cmp = self.object_id <=> other.object_id if cmp == 0
+      return cmp
     end
   end
 end

data/lib/hierclust/version.rb

@@ -2,7 +2,7 @@ module Hierclust #:nodoc:
   module VERSION #:nodoc:
     MAJOR = 0
     MINOR = 1
-    TINY  = 1
+    TINY  = 2
 
     STRING = [MAJOR, MINOR, TINY].join('.')
   end

data/spec/hierclust/distances_spec.rb

@@ -11,14 +11,6 @@ module Hierclust
       @dist = Math.sqrt((@x_1 - @x_2) ** 2 + (@y_1 - @y_2) ** 2)
     end
 
-    it "should tell us the distance between those points" do
-      @d[@p_1, @p_2].should == @dist
-    end
-
-    it "should tell us the distance between those points backwards" do
-      @d[@p_1, @p_2].should == @dist
-    end
-
     it "should have separation equal to distance between those points" do
       @d.separation.should == @dist
     end

data/spec/hierclust/point_spec.rb

@@ -6,7 +6,8 @@ module Hierclust
       @p = Point.new(1, 1)
     end
 
-    it "should return no clusters" do
+    it "should have points that is an array of only self" do
+      @p.points.should == [@p]
     end
   end
 end

data/website/index.html

@@ -33,7 +33,7 @@
     <h1>Simple Hierarchical Clustering</h1>
     <div id="version" class="clickable" onclick='document.location = "http://rubyforge.org/projects/hierclust"; return false'>
       <p>Get Version</p>
-      <a href="http://rubyforge.org/projects/hierclust" class="numbers">0.1.1</a>
+      <a href="http://rubyforge.org/projects/hierclust" class="numbers">0.1.2</a>
     </div>
     <h1>&#x2192; &#8216;hierclust&#8217;</h1>
 
@@ -105,7 +105,7 @@ to your data:</p>
   <span class="ident">map</span><span class="punct">.</span><span class="ident">add_point</span><span class="punct">(</span>
     <span class="ident">x</span> <span class="punct">=&gt;</span> <span class="ident">cluster</span><span class="punct">.</span><span class="ident">x</span><span class="punct">,</span>
     <span class="ident">y</span> <span class="punct">=&gt;</span> <span class="ident">cluster</span><span class="punct">.</span><span class="ident">y</span><span class="punct">,</span>
-    <span class="ident">label</span> <span class="punct">=&gt;</span> <span class="punct">&quot;</span><span class="string"><span class="expr">#{cluster.size}</span> Things</span><span class="punct">&quot;</span>
+    <span class="ident">label</span> <span class="punct">=&gt;</span> <span class="punct">&quot;</span><span class="string"><span class="expr">#{cluster.points}</span> Things</span><span class="punct">&quot;</span>
   <span class="punct">)</span>
 <span class="keyword">end</span></pre></p>
 
@@ -125,10 +125,7 @@ to your data:</p>
 	<h2>Source code</h2>
 
 
-	<p>The trunk repository is <code>svn://rubyforge.org/var/svn/hierclust/trunk</code> for anonymous access.</p>
-
-
-	<p>You can also browse the source online at <a href="http://hierclust.rubyforge.org/svn/trunk/">http://hierclust.rubyforge.org/svn/trunk/</a></p>
+	<p>You can browse the source at <a href="http://hierclust.rubyforge.org/svn/trunk/">http://hierclust.rubyforge.org/svn/trunk/</a></p>
 
 
 	<h2>How to submit patches</h2>
@@ -148,7 +145,7 @@ to your data:</p>
 
 	<p>Comments are welcome. Send an email to <a href="mailto:brandt@kurowski.net">Brandt Kurowski</a> email via the <a href="http://groups.google.com/group/hierclust">forum</a></p>
     <p class="coda">
-      <a href="http://brandt.kurowski.net/">Brandt Kurowski</a>, 4th February 2008<br>
+      <a href="http://brandt.kurowski.net/">Brandt Kurowski</a>, 6th February 2008<br>
       Theme extended from <a href="http://rb2js.rubyforge.org/">Paul Battley</a>
     </p>
 </div>

data/website/index.txt

@@ -52,7 +52,7 @@ clusters.each do |cluster|
   map.add_point(
     x => cluster.x,
     y => cluster.y,
-    label => "#{cluster.size} Things"
+    label => "#{cluster.points} Things"
   )
 end</pre>
 
@@ -66,9 +66,7 @@ h2. Forum
 
 h2. Source code
 
-The trunk repository is <code>svn://rubyforge.org/var/svn/hierclust/trunk</code> for anonymous access.
-
-You can also browse the source online at "http://hierclust.rubyforge.org/svn/trunk/":http://hierclust.rubyforge.org/svn/trunk/
+You can browse the source at "http://hierclust.rubyforge.org/svn/trunk/":http://hierclust.rubyforge.org/svn/trunk/
 
 h2. How to submit patches
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: hierclust
 version: !ruby/object:Gem::Version 
-  version: 0.1.1
+  version: 0.1.2
 platform: ruby
 authors: 
 - Brandt Kurowski
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-02-04 00:00:00 -05:00
+date: 2008-02-08 00:00:00 -05:00
 default_executable: 
 dependencies: []