fixed: percent float instead of int, sum of number counts for cv

1 files changed, 24 insertions, 12 deletions

diff --git a/lib/validation_db.rb b/lib/validation_db.rb
index 01607ce..8f4a540 100644
--- a/lib/validation_db.rb
+++ b/lib/validation_db.rb
@@ -3,33 +3,45 @@
   require lib
 end
 
+require "lib/merge.rb"
+
 module Lib
 
-  VAL_PROPS = [ :id, :uri, :model_uri, :training_dataset_uri, :prediction_feature,
-                :test_dataset_uri, :prediction_dataset_uri,  
-                :created_at, :real_runtime, # :cpu_runtime, 
-                :num_instances, :num_without_class, :percent_without_class, :num_unpredicted, :percent_unpredicted ] 
+  VAL_PROPS_GENERAL = [ :id, :uri, :model_uri, :training_dataset_uri, :prediction_feature,
+                :test_dataset_uri, :prediction_dataset_uri, :created_at ] 
+  VAL_PROPS_SUM = [ :num_instances, :num_without_class, :num_unpredicted ]
+  VAL_PROPS_AVG = [:real_runtime, :percent_without_class, :percent_unpredicted ]
+  VAL_PROPS = VAL_PROPS_GENERAL + VAL_PROPS_SUM + VAL_PROPS_AVG
   
   # :crossvalidation_info
   VAL_CV_PROPS = [ :crossvalidation_id, :crossvalidation_fold ]
   
   # :classification_statistics
-  VAL_CLASS_PROPS_SINGLE = [ :num_correct, :num_incorrect, :percent_correct, :percent_incorrect ]
+  VAL_CLASS_PROPS_SINGLE_SUM = [ :num_correct, :num_incorrect, :confusion_matrix  ]
+  VAL_CLASS_PROPS_SINGLE_AVG = [ :percent_correct, :percent_incorrect ]
+  VAL_CLASS_PROPS_SINGLE = VAL_CLASS_PROPS_SINGLE_SUM + VAL_CLASS_PROPS_SINGLE_AVG
+  
   # :class_value_statistics
-  VAL_CLASS_PROPS_PER_CLASS = [ :area_under_roc, :false_negative_rate, :false_positive_rate,
-                                :f_measure, :num_false_positives, :num_false_negatives, 
-                                :num_true_positives, :num_true_negatives, :precision, 
+  VAL_CLASS_PROPS_PER_CLASS_SUM = [ :num_false_positives, :num_false_negatives, 
+                                :num_true_positives, :num_true_negatives ]
+  VAL_CLASS_PROPS_PER_CLASS_AVG = [ :area_under_roc, :false_negative_rate, :false_positive_rate,
+                                :f_measure, :precision, 
                                 :recall, :true_negative_rate, :true_positive_rate ]
-  VAL_CLASS_PROPS = VAL_CLASS_PROPS_SINGLE + VAL_CLASS_PROPS_PER_CLASS + [ :confusion_matrix ]
+  VAL_CLASS_PROPS_PER_CLASS = VAL_CLASS_PROPS_PER_CLASS_SUM + VAL_CLASS_PROPS_PER_CLASS_AVG
+                                
+  VAL_CLASS_PROPS = VAL_CLASS_PROPS_SINGLE + VAL_CLASS_PROPS_PER_CLASS
   VAL_CLASS_PROPS_EXTENDED = VAL_CLASS_PROPS + [:accuracy]
 
   # :regression_statistics
   VAL_REGR_PROPS = [ :root_mean_squared_error, :mean_absolute_error, :r_square ]
-  
   CROSS_VAL_PROPS = [:algorithm_uri, :dataset_uri, :num_folds, :stratified, :random_seed]
   
   ALL_PROPS = VAL_PROPS + VAL_CV_PROPS + VAL_CLASS_PROPS_EXTENDED + VAL_REGR_PROPS + CROSS_VAL_PROPS
 
+  VAL_MERGE_GENERAL = VAL_PROPS_GENERAL + VAL_CV_PROPS + [:classification_statistics, :regression_statistics] + CROSS_VAL_PROPS
+  VAL_MERGE_SUM = VAL_PROPS_SUM + VAL_CLASS_PROPS_SINGLE_SUM + VAL_CLASS_PROPS_PER_CLASS_SUM
+  VAL_MERGE_AVG = VAL_PROPS_AVG + VAL_CLASS_PROPS_SINGLE_AVG + VAL_CLASS_PROPS_PER_CLASS_AVG + VAL_REGR_PROPS
+ 
   class Validation
     include DataMapper::Resource 
   
@@ -45,9 +57,9 @@ module Lib
     
     property :num_instances, Integer
     property :num_without_class, Integer
-    property :percent_without_class, Integer
+    property :percent_without_class, Float
     property :num_unpredicted, Integer
-    property :percent_unpredicted, Integer
+    property :percent_unpredicted, Float
         
     property :classification_statistics, Object #Hash
     property :regression_statistics, Object