Merged back to d7c0289 but includes fixes.

Moved back to setting with best performance (see table 18,
http://goo.gl/uGMJN).

3 files changed, 5 insertions, 39 deletions

diff --git a/lib/algorithm.rb b/lib/algorithm.rb
index 19666ad..ebd2019 100644
--- a/lib/algorithm.rb
+++ b/lib/algorithm.rb
@@ -517,16 +517,9 @@ module OpenTox
          
           # Recursive feature elimination
           rfProfile = rfe( x=features, y=y, rfeControl=rfeControl(functions=rfFuncs, number=150), sizes=subsets)
-          optVar = rfProfile$optVariables
-          if (rfProfile$bestSubset == dim(features)[2]) {
-            newRMSE = rfProfile$results$RMSE
-            newRMSE[which.min(rfProfile$results$RMSE)] = Inf
-            newOptSize = rfProfile$results[which.min(newRMSE),]$Variables
-            optVar = rfProfile$optVariables[1:newOptSize]
-          }
           
           # read existing dataset and select most useful features
-          csv=feats[,c("SMILES", optVar)]
+          csv=feats[,c("SMILES", rfProfile$optVariables)]
           write.csv(x=csv,file=f_fds_r, row.names=F, quote=F, na='')
         EOR
         r_result_file
diff --git a/lib/parser.rb b/lib/parser.rb
index f11ed48..e871323 100644
--- a/lib/parser.rb
+++ b/lib/parser.rb
@@ -353,11 +353,11 @@ module OpenTox
       # @param [Boolean] all_numeric Whether all features should be treated as numeric
       # @param [Boolean] del_nominal All nominal features will be removed
       # @return [OpenTox::Dataset] Dataset object with CSV data
-      def load_csv(csv, drop_missing=false, all_numeric=false, del_nominal=false)
-        
+      def load_csv(csv, drop_missing=false, all_numeric=false)
         row = 0
         input = csv.split("\n")
         headers = split_row(input.shift)
+        add_features(headers)
         value_maps = Array.new
         regression_features=Array.new
 
@@ -372,29 +372,6 @@ module OpenTox
             end
           }
         }
-        
-        if del_nominal
-          # Collect all nominal features 
-          del_features = Array.new
-          regression_features.each_with_index {|value, i|
-            if value == false
-              del_features << i
-            end
-          }
-          del_features = del_features.reverse
-          # Delete nominal entries in each row
-          del_features.each{|f|
-            regression_features.delete_at(f)
-            headers.delete_at(f+1)
-            input.map!{|row|
-              new_row = split_row(row)
-              new_row.delete_at(f+1)
-              row = new_row.join(",")
-            }
-          }
-        end
-        
-        add_features(headers)
 
         input.each_with_index { |row, i| 
           drop=false
diff --git a/lib/utils.rb b/lib/utils.rb
index eccec46..40988db 100644
--- a/lib/utils.rb
+++ b/lib/utils.rb
@@ -55,16 +55,12 @@ module OpenTox
           master = jl_master if jl_master
           master = ambit_master if ambit_master
         end
-        
+
         parser = OpenTox::Parser::Spreadsheets.new
         ds = OpenTox::Dataset.new
         ds.save
         parser.dataset = ds
-        if compounds.size < 4 
-          ds = parser.load_csv(master.collect{|r| r.join(",")}.join("\n"),false,true,false) # all_numeric = true, all features should be treated as numeric
-        else
-          ds = parser.load_csv(master.collect{|r| r.join(",")}.join("\n"),false,false,true) # del_nominal = true, which removes nominal features
-        end
+        ds = parser.load_csv(master.collect{|r| r.join(",")}.join("\n"),false,true)
         ds.save
 
       rescue Exception => e