add contra splitting, provide access to dataset-splits, minor stuff

3 files changed, 70 insertions, 17 deletions

diff --git a/validation/validation_application.rb b/validation/validation_application.rb
index 9233502..c873a72 100755
--- a/validation/validation_application.rb
+++ b/validation/validation_application.rb
@@ -9,8 +9,8 @@ require 'validation/validation_service.rb'
 helpers do
   def check_stratified(params)
     params[:stratified] = "false" unless params[:stratified]
-    raise OpenTox::BadRequestError.new "stratified != true|false|super|super4|super5|anti, is #{params[:stratified]}" unless
-      params[:stratified]=~/^(true|false|super|super4|super5|anti)$/
+    raise OpenTox::BadRequestError.new "stratified != true|false|super|super4|super5|contra, is #{params[:stratified]}" unless
+      params[:stratified]=~/^(true|false|super|super4|super5|contra)$/
   end
 end
 
@@ -745,6 +745,47 @@ end
 #  return validation.send(params[:attribute])
 #end
 
+
+def get_splits(id)
+  require "#{ENV['HOME']}/workspace/ValidationExperiments/dataset_split.rb"
+  validation = Validation::Validation[id]
+  raise OpenTox::NotFoundError.new "Validation '#{id}' not found." unless validation
+  Exp::DatasetSplit.find({:train_dataset_uri => validation.training_dataset_uri,:test_dataset_uri => validation.test_dataset_uri,})
+end
+
+def get_split(id,id2)
+  get_splits(id).each do |s|
+    return s if id2.to_s==s.id.to_s
+  end
+  raise "not found: dataset split with id #{id2}" unless split 
+end
+
+get '/:id/split' do
+  splits = get_splits(params[:id])
+  base_uri = "http://local-ot/validation/#{params[:id]}/split/"
+  uris = splits.collect{|s| base_uri+s.id}
+  if request.env['HTTP_ACCEPT'] =~ /text\/html/
+    content_type "text/html"
+    OpenTox.text_to_html uris.join("\n")
+  else
+    content_type "text/uri-list"
+    uris.join("\n")
+  end
+end
+
+get '/:id/split/:id2' do
+  split = get_split(params[:id],params[:id2])
+  content_type "text/html"
+  split.inspect
+  OpenTox.text_to_html ["http://local-ot/validation/#{params[:id]}/split/#{params[:id2]}/viz",split].to_yaml
+end
+
+get '/:id/split/:id2/viz' do
+  split = get_split(params[:id],params[:id2])
+  content_type("image/svg+xml")
+  result = body(File.new(split.svg_path))
+end
+
 get '/:id/migrate_median_confidence' do
   LOGGER.debug "migrate median confidence"
   validation = Validation::Validation[params[:id]]
diff --git a/validation/validation_service.rb b/validation/validation_service.rb
index e099e25..f57b3a5 100755
--- a/validation/validation_service.rb
+++ b/validation/validation_service.rb
@@ -653,15 +653,21 @@ module Validation
       meta = { DC.creator => $url_provider.url_for('/training_test_split',:full) }
       
       case stratified
-      when /^(true|super|super4|super5|anti)$/
+      when /^(true|super|super4|super5|contra)$/
         raise "store split clusters not available for true stratified splits" if store_split_clusters and stratified=="true"
         if stratified=="true"
           raise OpenTox::BadRequestError.new "prediction feature required for stratified splits" unless prediction_feature
           LOGGER.warn "split features are ignored for stratified splits (use super instead)" if features
           features = [prediction_feature]
+        elsif stratified=="contra"
+          raise OpenTox::BadRequestError.new "prediction feature required for anti-stratified splits" unless prediction_feature
+          LOGGER.debug "prediction feature is removed for anti-stratified splits"
+          features = orig_dataset.features.keys-[prediction_feature] 
+          raise unless features.size==orig_dataset.features.size-1
         else
           LOGGER.warn "prediction feature is ignored for super- or anti-stratified splits" if prediction_feature
         end
+        LOGGER.debug "Using "+features.size.to_s+"/"+orig_dataset.features.size.to_s+" features for splitting" if features
         r_util = OpenTox::RUtil.new 
         train, test = r_util.stratified_split( orig_dataset, meta, missing_values, split_ratio, 
           @subjectid, random_seed, features, stratified, store_split_clusters )
@@ -705,7 +711,7 @@ module Validation
         end
         LOGGER.debug "split done, training dataset: '"+result[:training_dataset_uri].to_s+"', test dataset: '"+result[:test_dataset_uri].to_s+"'"
       else
-        raise OpenTox::BadRequestError.new "stratified != false|true|super, is #{stratified}"
+        raise OpenTox::BadRequestError.new "stratified != false|true|super|contra, is #{stratified}"
       end
       result
     end
diff --git a/validation/validation_test.rb b/validation/validation_test.rb
index 1a41881..a7fae88 100755
--- a/validation/validation_test.rb
+++ b/validation/validation_test.rb
@@ -60,6 +60,10 @@ class ValidationTest < Test::Unit::TestCase
     begin
       $test_case = self
       
+      get '2568/split'
+      puts last_response.body
+      exit
+      
 #      dataset_uri = "http://apps.ideaconsult.net:8080/ambit2/dataset/603206?pagesize=250&page=0"
 #      test_dataset_uri = "http://apps.ideaconsult.net:8080/ambit2/dataset/603206?pagesize=250&page=1"
 #      #prediction_feature = "http://apps.ideaconsult.net:8080/ambit2/feature/528321"
@@ -85,16 +89,16 @@ class ValidationTest < Test::Unit::TestCase
 #        {:validation_uris=>"http://local-ot/validation/389,http://local-ot/validation/390,http://local-ot/validation/391,http://local-ot/validation/392",
 #         :identifier=>"split1,split1,split2,split2"}
       
-      post "/training_test_validation",{:prediction_feature=>"http://local-ot/dataset/9264/feature/endpoint", 
-        :training_dataset_uri=>"http://local-ot/dataset/119127", 
-        :algorithm_uri=>"http://local-ot/weka/RandomForest", 
-        :test_dataset_uri=>"http://local-ot/dataset/119128",
-        :test_target_dataset_uri=>"http://local-ot/dataset/9264",  
-        :algorithm_params=>""}
-      uri = last_response.body
-      rep = wait_for_task(uri)
-      puts rep
-      exit
+#      post "/training_test_validation",{:prediction_feature=>"http://local-ot/dataset/9264/feature/endpoint", 
+#        :training_dataset_uri=>"http://local-ot/dataset/119127", 
+#        :algorithm_uri=>"http://local-ot/weka/RandomForest", 
+#        :test_dataset_uri=>"http://local-ot/dataset/119128",
+#        :test_target_dataset_uri=>"http://local-ot/dataset/9264",  
+#        :algorithm_params=>""}
+#      uri = last_response.body
+#      rep = wait_for_task(uri)
+#      puts rep
+#      exit
 #        
 #        #post "/report/validation",{:validation_uris=>"http://local-ot/validation/22849",:min_confidence=>0.5}
 #      get "/22849",{:min_confidence=>0.5}
@@ -108,10 +112,12 @@ class ValidationTest < Test::Unit::TestCase
 #      #dataset_uri = "http://apps.ideaconsult.net:8080/ambit2/dataset/603306?feature_uris[]=http://apps.ideaconsult.net:8080/ambit2/feature/764036"
 #      #dataset_uri = "http://apps.ideaconsult.net:8080/ambit2/dataset/603204"
       
-      #dataset_uri = "http://local-ot/dataset/9264"
-      #post "/plain_training_test_split",{:dataset_uri=>dataset_uri, :stratified=>"super", :split_ratio=>0.25}
+      dataset_uri = "http://local-ot/dataset/12084"
+      prediction_feature = "http://local-ot/dataset/12084/feature/bbrc/1"
+      post "/plain_training_test_split",{:dataset_uri=>dataset_uri, :stratified=>"contra", :split_ratio=>0.25, :missing_values=>"0",
+        :prediction_feature => prediction_feature}
       
-      get '31355/median_confidence'
+      #get '31355/median_confidence'
       puts last_response.body
       exit