lite 0.0.3 → 0.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 6d79082f5ca6160a1ef4b0356e6708b22fe0d67a
-  data.tar.gz: 0a05a1776c11c33497a1f9afcbc29d5eca1f5912
+  metadata.gz: 5be315e570f8fa335d6059e3e1c03b168618d6fc
+  data.tar.gz: 65d024a05f399a503c094f994b4515651a0cfd86
 SHA512:
-  metadata.gz: 4445ce29ed39fe1da5cfed1b9825c4d8a2ced615939c27d4f02c77e89219c2f240e72b6a8c5202246a2317352228faffefb80d3729664038ba8cbf134b92181f
-  data.tar.gz: 6cbf5e1f884c1ff9a7d52f4a10d4e5aacf0bdc54b1b12e7dac3e5f8b2bf07129664dd84ed7ecd0ceb0e4afa00178979c89a5ec5e0fa20f9359b6f22966024529
+  metadata.gz: 0eabf693fbfbc5233bd63cf0eaa1b502c4ae1eafa47170fd00a5c87201d1b168ba713f5ccbe3e49e3bcb7abdb22a9ab94caec8edb291a4d45c3cd37a43fb3d8f
+  data.tar.gz: 3ada7a12747ae9cc62d9326ab75cfcadec2a297f86fae466e6cfbc24e48d186e50b1b44e55a5ce527adf311df3bad37e8e3807a062bb6418e75aa5d94f59beaa

data/lib/lite/cluster.rb

@@ -8,15 +8,16 @@ module Cluster
     end
 
     def observe!( instance )
-      cinstance = Centroid.new( instance )
+      u = SparseVector.new instance
+
       if @centroids.size == 0
-        @centroids << cinstance
+        @centroids << Centroid.new( u )
         return self
       end
 
-      @centroids.sort! {|c1, c2|   instance.dist(c1.x) <=> instance.dist(c2.x) }
+      @centroids.sort! {|c1, c2|  u.dist(c1.x) <=> u.dist(c2.x) }
       closest_centroid = @centroids.first
-      closest_centroid.update!( cinstance )
+      closest_centroid.update!( u )
 
       if( @centroids.size >= @k_max ) 
         pairs = [] 
@@ -36,15 +37,14 @@ module Cluster
         @centroids[merge_info[1]].merge!( @centroids[merge_info[2]] )
         @centroids = @centroids - [ @centroids[merge_info[2]] ]
       end
-      @centroids << Centroid.new( cinstance )
+      @centroids << Centroid.new( u )
       
       []
     end
 
-    def getCentroids( min_num_instances_in_cluster = 2 )      
+    def centroids( min_num_instances_in_cluster = 2 )      
       @centroids.each do |c|        
         next if c.n >= min_num_instances_in_cluster
-        p c
         @centroids = @centroids - [ c ]      
         next if c.n == 0
         aux = @centroids.inject( {:min_c => @centroids.first, :d => @centroids.first.x.dist( c.x )} ) {|a,cc| cc.nil? || cc.x.dist(c.x) > a[:d] ? a : { :min_c=>cc, :d=>cc.x.dist(c.x)} }        
@@ -54,15 +54,15 @@ module Cluster
     end
   end
 
-  class Centroid < SparseVector
-    attr_accessor :x,:n
+  class Centroid
+    attr_accessor :n,:x
     def initialize( x )
       @x = x
-      @n = 0
+      @n = 1
     end
 
     def update!( newX )
-      @x += ( @x - newX ).mult_scalar( 1.0/(@n+1) )
+      @x += ( @x.mult_scalar(@n) + newX ).mult_scalar( 1.0/(@n+1) )
       @n +=  1
       self 
     end
@@ -72,6 +72,10 @@ module Cluster
       @n += centroid.n
       self
     end
+
+    def to_s
+      "Centroid #{x.attr} of #{n} instances"
+    end
   end
  
 end 

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: lite
 version: !ruby/object:Gem::Version
-  version: 0.0.3
+  version: 0.0.4
 platform: ruby
 authors:
 - ronbee