groupie_spec.rb in groupie-0.2.2

- old
+ new

@@ -43,11 +43,41 @@
       c = g.classify('discreetly')
       c[:spam].should > c[:ham]
       c2 = g.classify('user')
       c2[:ham].should > c2[:spam]
     end
+
+    describe "strategies" do
+      describe "sum" do
+        it "should weigh words for the sum of their occurances" do
+          g = Groupie.new
+          g[:spam].add %w[word] * 9
+          g[:ham].add %w[word]
+          g.classify('word', :sum).should == {:spam=>0.9, :ham=>0.1}
+        end
+      end
+
+      describe "sqrt" do
+        it "should weigh words for the square root of the sum of ocurances" do
+          g = Groupie.new
+          g[:spam].add %w[word] * 9
+          g[:ham].add %w[word]
+          g.classify('word', :sqrt).should == {:spam=>0.75, :ham=>0.25}
+        end
+      end
+
+      describe "log" do
+        it "should weigh words for log10 of their sum of occurances" do
+          g = Groupie.new
+          g[:spam].add %w[word] * 1000
+          g[:ham].add %w[word] * 10
+          g.classify('word', :log).should == {:spam=>0.75, :ham=>0.25}
+        end
+      end
+    end
   end
+
   context "classify_text" do
     it 'should tokenized html emails' do
       g = Groupie.new
       spam_tokens = File.read(File.join(File.dirname(__FILE__), %w[fixtures spam spam.la-44118014.txt])).tokenize
       ham_tokens = File.read(File.join(File.dirname(__FILE__), %w[fixtures ham spam.la-44116217.txt])).tokenize
@@ -68,8 +98,33 @@
       g[:ham].add %w[buy flowers for your mom]
       result = g.classify_text "Grow flowers to sell on our website".tokenize
       result[:spam].should > result[:ham]
       result2 = g.classify_text "Grow flowers to give to your mom".tokenize
       result2[:ham].should == result2[:spam]
+    end
+
+    it "should skip unknown tokens" do
+      g = Groupie.new
+      g[:spam].add %w[buy viagra now]
+      g[:ham].add %w[buy flowers now]
+      g.classify_text(%w[buy buckets now]).should == {:spam=>0.5, :ham=>0.5}
+    end
+    
+    it "should support the sqrt strategy" do
+      g = Groupie.new
+      g[:spam].add %w[one] * 9
+      g[:ham].add %w[one]
+      g[:spam].add %w[two] * 9
+      g[:ham].add %w[two]
+      g.classify_text(%w[one two three], :sqrt).should == {:spam=>0.75, :ham=>0.25}
+    end
+    
+    it "should support the log strategy" do
+      g = Groupie.new
+      g[:spam].add %w[one] * 100
+      g[:ham].add %w[one]
+      g[:spam].add %w[two]
+      g[:ham].add %w[two] * 100
+      g.classify_text(%w[one two three], :log).should == {:spam=>0.5, :ham=>0.5}
     end
   end
 end