probabilities branch merged

5 files changed, 30 insertions, 29 deletions

diff --git a/VERSION b/VERSION
index 2003b63..965065d 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-0.9.2
+0.9.3
diff --git a/ext/lazar/extconf.rb b/ext/lazar/extconf.rb
index 0e607f3..d3d2756 100644
--- a/ext/lazar/extconf.rb
+++ b/ext/lazar/extconf.rb
@@ -20,7 +20,9 @@ rinstall = File.expand_path(File.join(File.dirname(__FILE__),"rinstall.R"))
 puts `Rscript --vanilla #{rinstall} #{r_dir}`
 
 r_libs = Dir[File.join(r_dir,"*")].collect{|l| l.sub(r_dir, '').sub('/','')}.sort
-abort "Failed to install R packages." unless r_libs == ["caret","doMC","foreach","ggplot2","gridExtra","iterators","pls"].sort
+["caret","doMC","foreach","ggplot2","gridExtra","iterators","pls"].each do |lib|
+  abort "Failed to install R package '#{lib}'." unless r_libs.include?(lib)
+end
 
 # create a fake Makefile
 File.open(File.join(File.dirname(__FILE__),"Makefile"),"w+") do |makefile|
diff --git a/lib/classification.rb b/lib/classification.rb
index 0de8726..93b4f0f 100644
--- a/lib/classification.rb
+++ b/lib/classification.rb
@@ -5,29 +5,27 @@ module OpenTox
 
       def self.weighted_majority_vote compound, params
         neighbors = params[:neighbors]
-        weighted_sum = {}
-        sim_sum = 0.0
-        confidence = 0.0
-        # see ~/src/pubchem-read-across/application.rb:353
-        neighbors.each do |row|
-          sim = row["tanimoto"]
-          row["toxicities"][params[:prediction_feature_id].to_s].each do |act|
-            weighted_sum[act] ||= 0
-            weighted_sum[act] += sim
+        feature_id = params[:prediction_feature_id].to_s
+        sims = {}
+        neighbors.each do |n|
+          sim = n["tanimoto"]
+          n["toxicities"][feature_id].each do |act|
+            sims[act] ||= []
+            sims[act] << sim
           end
         end
-        case weighted_sum.size
-        when 1
-          return {:value => weighted_sum.keys.first, :confidence => weighted_sum.values.first/neighbors.size.abs}
-        when 2
-          sim_sum = weighted_sum[weighted_sum.keys[0]]
-          sim_sum -= weighted_sum[weighted_sum.keys[1]]
-          sim_sum > 0 ? prediction = weighted_sum.keys[0] : prediction = weighted_sum.keys[1] 
-          confidence = (sim_sum/neighbors.size).abs 
-          return {:value => prediction,:confidence => confidence}
-        else
-          bad_request_error "Cannot predict more than 2 classes, multinomial classifications is not yet implemented. Received classes were: '#{weighted.sum.keys}'"
+        sim_all = sims.collect{|a,s| s}.flatten
+        sim_sum = sim_all.sum
+        sim_max = sim_all.max
+        probabilities = {}
+        sims.each do |a,s|
+          probabilities[a] = s.sum/sim_sum
         end
+        probabilities = probabilities.collect{|a,p| [a,sim_max*p]}.to_h
+        p_max = probabilities.collect{|a,p| p}.max
+        prediction = probabilities.key(p_max)
+        {:value => prediction,:probabilities => probabilities}
+
       end
     end
   end
diff --git a/lib/crossvalidation.rb b/lib/crossvalidation.rb
index 752d393..50afb6f 100644
--- a/lib/crossvalidation.rb
+++ b/lib/crossvalidation.rb
@@ -54,7 +54,7 @@ module OpenTox
       cv.update_attributes(
         nr_instances: nr_instances,
         nr_unpredicted: nr_unpredicted,
-        predictions: predictions#.sort{|a,b| b[3] <=> a[3]} # sort according to confidence
+        predictions: predictions
       )
       $logger.debug "Nr unpredicted: #{nr_unpredicted}"
       cv.statistics
diff --git a/lib/validation-statistics.rb b/lib/validation-statistics.rb
index 570b2d4..c6b2a07 100644
--- a/lib/validation-statistics.rb
+++ b/lib/validation-statistics.rb
@@ -15,21 +15,21 @@ module OpenTox
           if pred[:value] == m
             if pred[:value] == accept_values[0]
               confusion_matrix[0][0] += 1
-              weighted_confusion_matrix[0][0] += pred[:confidence]
+              weighted_confusion_matrix[0][0] += pred[:probabilities][pred[:value]]
               nr_instances += 1
             elsif pred[:value] == accept_values[1]
               confusion_matrix[1][1] += 1
-              weighted_confusion_matrix[1][1] += pred[:confidence]
+              weighted_confusion_matrix[1][1] += pred[:probabilities][pred[:value]]
               nr_instances += 1
             end
           elsif pred[:value] != m
             if pred[:value] == accept_values[0]
               confusion_matrix[0][1] += 1
-              weighted_confusion_matrix[0][1] += pred[:confidence]
+              weighted_confusion_matrix[0][1] += pred[:probabilities][pred[:value]]
               nr_instances += 1
             elsif pred[:value] == accept_values[1]
               confusion_matrix[1][0] += 1
-              weighted_confusion_matrix[1][0] += pred[:confidence]
+              weighted_confusion_matrix[1][0] += pred[:probabilities][pred[:value]]
               nr_instances += 1
             end
           end
@@ -47,14 +47,15 @@ module OpenTox
           confidence_sum += c
         end
       end
-      accuracy = (weighted_confusion_matrix[0][0]+weighted_confusion_matrix[1][1])/confidence_sum.to_f
+      accuracy = (confusion_matrix[0][0]+confusion_matrix[1][1])/nr_instances.to_f
+      weighted_accuracy = (weighted_confusion_matrix[0][0]+weighted_confusion_matrix[1][1])/confidence_sum.to_f
       $logger.debug "Accuracy #{accuracy}"
       {
         :accept_values => accept_values,
         :confusion_matrix => confusion_matrix,
         :weighted_confusion_matrix => weighted_confusion_matrix,
         :accuracy => accuracy,
-        :weighted_accuracy => (weighted_confusion_matrix[0][0]+weighted_confusion_matrix[1][1])/confidence_sum.to_f,
+        :weighted_accuracy => weighted_accuracy,
         :true_rate => true_rate,
         :predictivity => predictivity,
         :finished_at => Time.now