lib/bayes/fisher.rb in yanbi-ml-0.2.1 vs lib/bayes/fisher.rb in yanbi-ml-0.2.2

@@ -3,20 +3,14 @@
 # License::   MIT
 
 module Yanbi
 
   class Fisher < Bayes
-  
-    def classify(text)
-      max_score(text) do |cat, doc|
-        fisher_score(cat, doc)
-      end
-    end
-  
+   
     private
   
-    def fisher_score(category, document)
+    def score(category, document)
       features = document.words.uniq
       probs = features.map {|x| weighted_prob(x, category)}
       pscores = probs.reduce(&:*)
       score = -2 * Math.log(pscores)
       invchi2(score, features.count * 2)
@@ -28,12 +22,11 @@
       return 0 if sum.zero?
       wp / sum
     end
 
     def word_prob(cat, word)
-      all_word_count = @category_counts[cat].values.reduce(&:+)
       count = @category_counts[cat].has_key?(word) ? @category_counts[cat][word].to_f : 0 
-      count / all_word_count
+      count / @category_sizes[cat]
     end
 
     def weighted_prob(word, category, basicprob=nil, weight=1.0, ap=0.5)
       basicprob = category_prob(category, word)
       totals = @category_counts.inject(0) {|sum, cat| sum += cat.last[word].to_i}