hierclust 0.1.3 → 0.1.4

data/History.txt

@@ -1,3 +1,10 @@
+== 0.1.4 2008-02-13
+
+* 1 minor enhancement:
+  * gave linear-time preclustering an independent "resolution" parameter
+* 1 bugfix:
+  * corrected cluster coordinate calculation
+
 == 0.1.3 2008-02-10
 
 * 1 performance improvement

data/lib/hierclust/cluster.rb

@@ -15,13 +15,15 @@ module Hierclust
     # Returns the average x-coordinates of all items in this Cluster.
     def x
       return nil if size == 0
-      @x ||= @items.inject(0) {|sum, p| sum + p.x} / size
+      points = self.points
+      @x ||= points.inject(0.0) {|sum, p| sum + p.x} / points.size
     end
 
     # Returns the average y-coordinates of all items in this Cluster.
     def y
       return nil if size == 0
-      @y ||= @items.inject(0) {|sum, p| sum + p.y} / size
+      points = self.points
+      @y ||= points.inject(0.0) {|sum, p| sum + p.y} / points.size
     end
     
     # Add an +item+ to this Cluster.

data/lib/hierclust/clusterer.rb

@@ -10,8 +10,15 @@ module Hierclust
     # 
     # Specify +separation+ to stop the clustering process once all the
     # items are at least +separation+ units apart.
-    def initialize(data, separation = nil)
+    # 
+    # Specify +resolution+ to give a minimum size for clusters. Points that
+    # are within this distance from each other will not be hierarchically
+    # 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
       @data = precluster(data)
       @distances = Distances.new(@data)
     end
@@ -45,6 +52,10 @@ module Hierclust
     end
     
     def precluster(points)
+      if @resolution.nil?
+        # preclustering is only applicable given lower bound on resolution
+        return points.dup
+      end
       if @separation.nil?
         # can't precluster w/ no min separation given
         return points.dup
@@ -53,7 +64,7 @@ module Hierclust
         # if no separation is asked for, it's all one cluster
         return [Cluster.new(points)]
       end
-      grid_size = @separation / 2.0
+      grid_size = @resolution
       grid_clusters = Hash.new
       points.each do |point|
         grid_x = (point.x / grid_size).floor

data/lib/hierclust/version.rb

@@ -2,7 +2,7 @@ module Hierclust #:nodoc:
   module VERSION #:nodoc:
     MAJOR = 0
     MINOR = 1
-    TINY  = 3
+    TINY  = 4
 
     STRING = [MAJOR, MINOR, TINY].join('.')
   end

data/spec/hierclust/cluster_spec.rb

@@ -58,4 +58,32 @@ module Hierclust
       @points.should include(@p_1, @p_2)
     end
   end
+
+  describe Cluster, " with one point and one cluster" do
+    before do
+      @x_1, @x_2, @x_3 = 1, 2, 3
+      @y_1, @y_2, @y_3 = 2, 2, 5
+      @p_1 = Point.new(@x_1, @y_1)
+      @p_2 = Point.new(@x_2, @y_2)
+      @p_3 = Point.new(@x_3, @y_3)
+      @c_1 = Cluster.new([@p_1, @p_2])
+      @c_2 = Cluster.new([@p_3, @c_1])
+    end
+
+    it "should have two items" do
+      @c_2.items.size.should == 2
+    end
+
+    it "should have three points" do
+      @c_2.points.size.should == 3
+    end
+
+    it "should have x-coordinate at average of point's x-coordinates" do
+      @c_2.x.should == 2
+    end
+
+    it "should have y-coordinate at average of point's y-coordinates" do
+      @c_2.y.should == 3
+    end
+  end
 end

data/spec/hierclust/clusterer_spec.rb

@@ -164,18 +164,43 @@ module Hierclust
       end
 
       describe "and separation 3" do
-        before do
-          @clusters = Clusterer.new(@points, 3).clusters.sort
+        describe "with no resolution limit" do
+          before do
+            @clusters = Clusterer.new(@points, 3).clusters.sort
+          end
+
+          it "should have three clusters" do
+            @clusters.size.should == 3
+          end
+
+          it "should have clusters size 2, 2, and 4 " do
+            @clusters[0].points.size.should == 2
+            @clusters[1].points.size.should == 2
+            @clusters[2].points.size.should == 4
+          end
+
+          it "should have 2 items in large cluster" do
+            @clusters[2].items.size.should == 2
+          end
         end
 
-        it "should have three clusters" do
-          @clusters.size.should == 3
-        end
+        describe "with coarse resolution" do
+          before do
+            @clusters = Clusterer.new(@points, 3, 5).clusters.sort
+          end
+
+          it "should have three clusters" do
+            @clusters.size.should == 2
+          end
+
+          it "should have clusters size 2, 2, and 4 " do
+            @clusters[0].points.size.should == 4
+            @clusters[1].points.size.should == 4
+          end
 
-        it "should have clusters size 2, 2, and 4 " do
-          @clusters[0].points.size.should == 2
-          @clusters[1].points.size.should == 2
-          @clusters[2].points.size.should == 4
+          it "should have 4 items in large cluster" do
+            @clusters[1].items.size.should == 4
+          end
         end
       end
     end

data/spec/hierclust/point_spec.rb

@@ -4,10 +4,15 @@ module Hierclust
   describe Point do
     before do
       @p = Point.new(1, 1)
+      @p2 = Point.new(2, 1)
     end
 
     it "should have points that is an array of only self" do
       @p.points.should == [@p]
     end
+
+    it "should calculate distance to other" do
+      @p.distance_to(@p2).should == 1.0
+    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">0.1.3</a>
+      <a href="http://rubyforge.org/projects/hierclust" class="numbers">0.1.4</a>
     </div>
     <h1>&#x2192; &#8216;hierclust&#8217;</h1>
 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: hierclust
 version: !ruby/object:Gem::Version 
-  version: 0.1.3
+  version: 0.1.4
 platform: ruby
 authors: 
 - Brandt Kurowski
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-02-10 00:00:00 -05:00
+date: 2008-02-13 00:00:00 -05:00
 default_executable: 
 dependencies: []