adding weighted_accuracy, change accuracy computation, some minor changes

3 files changed, 23 insertions, 5 deletions

diff --git a/lib/ot_predictions.rb b/lib/ot_predictions.rb
index e28f35b..b0e9720 100755
--- a/lib/ot_predictions.rb
+++ b/lib/ot_predictions.rb
@@ -165,7 +165,9 @@ module Lib
       end
       raise "no array" unless v.is_a?(Array)
       if v.size>1
-        raise "multiple values"
+        v.uniq!
+        raise "not yet implemented: multiple non-equal values "+compound.to_s+" "+v.inspect if v.size>1
+        v = v[0]
       elsif v.size==1
         v = v[0]
       else
diff --git a/lib/predictions.rb b/lib/predictions.rb
index e73dda6..5850024 100755
--- a/lib/predictions.rb
+++ b/lib/predictions.rb
@@ -163,19 +163,36 @@ module Lib
     def percent_correct
       raise "no classification" unless @feature_type=="classification"
       return 0 if @num_with_actual_value==0
-      return 100 * @num_correct / @num_with_actual_value.to_f
+      return 100 * @num_correct / (@num_with_actual_value - @num_unpredicted).to_f
     end
     
     def percent_incorrect
       raise "no classification" unless @feature_type=="classification"
       return 0 if @num_with_actual_value==0
-      return 100 * @num_incorrect / @num_with_actual_value.to_f
+      return 100 * @num_incorrect / (@num_with_actual_value - @num_unpredicted).to_f
     end
     
     def accuracy
       return percent_correct / 100.0
     end
     
+    def weighted_accuracy
+      raise "no classification" unless @feature_type=="classification"
+      total = 0
+      correct = 0
+      (0..@predicted_values.size-1).each do |i|
+        if @predicted_values[i]!=nil
+          total += @confidence_values[i]
+          correct += @confidence_values[i] if @actual_values[i]==@predicted_values[i]
+        end
+      end
+      if total==0 || correct == 0
+        return 0  
+      else
+        return correct / total 
+      end
+    end
+
     def percent_unpredicted
       return 0 if @num_with_actual_value==0
       return 100 * @num_unpredicted / @num_with_actual_value.to_f
diff --git a/lib/validation_db.rb b/lib/validation_db.rb
index 4b852f9..c4cb2a2 100755
--- a/lib/validation_db.rb
+++ b/lib/validation_db.rb
@@ -18,10 +18,9 @@ module Lib
   # :classification_statistics
   VAL_CLASS_PROPS_SINGLE_SUM = [ :num_correct, :num_incorrect, :confusion_matrix  ]
   VAL_CLASS_PROPS_SINGLE_AVG = [ :percent_correct, :percent_incorrect, 
-    :weighted_area_under_roc, :accuracy ] 
+    :weighted_area_under_roc, :accuracy, :weighted_accuracy ] 
   VAL_CLASS_PROPS_SINGLE = VAL_CLASS_PROPS_SINGLE_SUM + VAL_CLASS_PROPS_SINGLE_AVG
   
-  
   # :class_value_statistics
   VAL_CLASS_PROPS_PER_CLASS_SUM = [ :num_false_positives, :num_false_negatives, 
                                 :num_true_positives, :num_true_negatives ]