new branch val_exp

4 files changed, 26 insertions, 10 deletions

diff --git a/lib/dataset_cache.rb b/lib/dataset_cache.rb
index 1af1d51..7a13e9b 100644
--- a/lib/dataset_cache.rb
+++ b/lib/dataset_cache.rb
@@ -12,6 +12,7 @@ module Lib
       return nil if (dataset_uri==nil)
       d = @@cache[dataset_uri.to_s+"_"+subjectid.to_s]
       if d==nil
+        LOGGER.debug "loading dataset #{dataset_uri}"
         d = OpenTox::Dataset.find(dataset_uri, subjectid)
         @@cache[dataset_uri.to_s+"_"+subjectid.to_s] = d
       end
diff --git a/lib/dataset_util.rb b/lib/dataset_util.rb
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/lib/dataset_util.rb
diff --git a/lib/prediction_data.rb b/lib/prediction_data.rb
index d387d24..5f11957 100644
--- a/lib/prediction_data.rb
+++ b/lib/prediction_data.rb
@@ -172,11 +172,16 @@ module Lib
         
         predicted_values = []
         confidence_values = []
+        dup_compounds = []
+        dup_actual_values = []
         count = 0
-        compounds.each do |c|
+        compounds.size.times do |i|
+          c = compounds[i]
           if prediction_dataset.compounds.index(c)==nil
             predicted_values << nil
             confidence_values << nil
+            dup_compounds << c
+            dup_actual_values << actual_values[i]
           else
             case feature_type
             when "classification"
@@ -184,19 +189,28 @@ module Lib
             when "regression"
               vals = regression_vals(prediction_dataset, c, predicted_variable)
             end
-            raise "not yet implemented: more than one prediction for one compound" if vals.size>1
-            predicted_values << vals[0]
-            if predicted_confidence
-              confidence_values << confidence_val(prediction_dataset, c, predicted_confidence)
-            else
-              confidence_values << nil
-            end
+            
+            #vals.uniq! #more than one prediciton is ok if it yields the equal value 
+            #raise "not yet implemented: more than one prediction for one compound '#{vals.inspect}'" if vals.size>1
+            #predicted_values << vals[0]
+            
+            vals.each do |val|
+              dup_compounds << c
+              dup_actual_values << actual_values[i]
+              predicted_values << val
+              
+              if predicted_confidence
+                confidence_values << confidence_val(prediction_dataset, c, predicted_confidence)
+              else
+                confidence_values << nil
+              end
+            end    
           end
           count += 1
         end
-        all_compounds += compounds
+        all_compounds += dup_compounds
         all_predicted_values += predicted_values
-        all_actual_values += actual_values
+        all_actual_values += dup_actual_values
         all_confidence_values += confidence_values
         
         task.progress( task_status += task_step ) if task # loaded predicted values and confidence
diff --git a/lib/validation_db.rb b/lib/validation_db.rb
index 086853e..17bd506 100755
--- a/lib/validation_db.rb
+++ b/lib/validation_db.rb
@@ -85,6 +85,7 @@ module Validation
     index :prediction_feature
     index :training_dataset_uri
     index :test_dataset_uri
+    index :finished
     
     attr_accessor :subjectid