Add argument to set dataset file path.

5 files changed, 61 insertions, 33 deletions

diff --git a/cv/data/utils/count_features.rb b/cv/data/utils/count_features.rb
index 1e272ac..c8a5893 100644
--- a/cv/data/utils/count_features.rb
+++ b/cv/data/utils/count_features.rb
@@ -4,7 +4,11 @@ require 'yaml'
 
 @subjectid = nil
 
-
+if ARGV.size != 1
+  puts "Args: path/to/dataset.yaml"
+  puts ARGV.size
+  exit
+end
 
 def count_features(ds_uri)
   puts ds_uri
@@ -13,9 +17,10 @@ def count_features(ds_uri)
 
   features = dataset.features.keys
   puts "# all features: #{features.size}"
-  
+  feature_names = []
   delete_features = []
   features.each{ |fn|
+    feature_names << fn.split("\/feature\/").last
     dataset.features[fn][RDF.type].each { |typestr|
       if typestr.include? "MissingFeature"
         delete_features << fn 
@@ -23,6 +28,9 @@ def count_features(ds_uri)
       end
     }
   }
+  @all_feature_names << feature_names.sort
+  @all_feature_names << "" 
+
   puts "# Missingfeatures: #{delete_features.size}"
   features = features - delete_features
   puts "# numeric features: #{features.size}"
@@ -31,18 +39,30 @@ end
 
 
 @missing_features = []
-
-ds = YAML::load_file("../datasets.yaml")
+@all_feature_names = []
+path = ARGV[0]
+puts path
+ds = YAML::load_file("#{path}")
+#ds = YAML::load_file("../datasets.yaml")
 ds.keys.each { |dataset|
   puts "----------"
   puts dataset
+  @all_feature_names << "" 
+  @all_feature_names << "------ new dataset ------" 
+  @all_feature_names << "-------- #{dataset} --------"
   ds[dataset].keys.each { |pc|
-    puts pc unless (pc == "dataset") || (pc == "test") || (pc == "training")
-    count_features(ds[dataset][pc]) unless (pc == "dataset") || (pc == "test") || (pc == "training") 
+    if !(pc == "dataset") || (pc == "test") || (pc == "training")
+      puts pc
+      @all_feature_names << "--- new feature: #{pc} ---" 
+      count_features(ds[dataset][pc])
+    end
   }
   puts "----------"
   puts
 }
 puts
 puts "Missing features over all datasets:"
-puts @missing_features.uniq!.to_yaml 
+puts @missing_features.uniq!.to_yaml
+puts 
+puts "All feature names:"
+puts @all_feature_names
diff --git a/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL.rb b/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL.rb
index bd86579..2442b01 100644
--- a/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL.rb
+++ b/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL.rb
@@ -2,15 +2,18 @@
 # # Author: Andreas Maunz, David Vorgrimmler
 # # @params: Dataset_name(see dataset_nestle.yaml), pc_type(electronic,cpsa or constitutional ... or nil to disable), prediction_algorithm(local_mlr_prop or local_svm_regression ...)
 
-if ARGV.size != 4 
-  puts "Args: ds_name, pc_type, algo, random_seed"
+if ARGV.size != 5 
+  puts "Args: ds_name, pc_type, algo, random_seed, path/to/dataset.yaml"
   puts ARGV.size
   exit
 end
 
-ds_file = "datasets.yaml"
-pwd=`pwd`
-path = "#{pwd.chop}/../data/#{ds_file}"
+#ds_file = "datasets.yaml"
+#pwd=`pwd`
+#path = "#{pwd.chop}/../data/#{ds_file}"
+path = ARGV[4]
+ds_file = path.split("/").last
+
 if File.exists?(path)
   puts "#{ds_file} exists"
 else
@@ -29,7 +32,7 @@ pc_type = ARGV[1] # e.g. electronic,cpsa or nil to disable
 algo = ARGV[2]    # e.g. local_svm_regression, local_mlr_prop
 r_seed = ARGV[3]  # 1, 2, ..., 10
 
-ds = YAML::load_file("../data/datasets.yaml")
+ds = YAML::load_file("#{path}")
 ds_uri = ds[ds_name]["dataset"]
 pc_ds_uri = ds[ds_name][pc_type]
 
diff --git a/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL_loo.rb b/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL_loo.rb
index 88adac2..34ec4f9 100644
--- a/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL_loo.rb
+++ b/cv/scripts/CV_ds_pctype_algo_rseed_LOAEL_loo.rb
@@ -2,15 +2,18 @@
 # # Author: Andreas Maunz, David Vorgrimmler
 # # @params: Dataset_name(see dataset_nestle.yaml), pc_type(electronic,cpsa or constitutional ... or nil to disable), prediction_algorithm(local_mlr_prop or local_svm_regression ...)
 
-if ARGV.size != 4 
-  puts "Args: ds_name, pc_type, algo, random_seed"
+if ARGV.size != 5 
+  puts "Args: ds_name, pc_type, algo, random_seed, path/to/dataset.yaml"
   puts ARGV.size
   exit
 end
 
-ds_file = "datasets.yaml"
-pwd=`pwd`
-path = "#{pwd.chop}/../data/#{ds_file}"
+#ds_file = "datasets.yaml"
+#pwd=`pwd`
+#path = "#{pwd.chop}/../data/#{ds_file}"
+path = ARGV[4]
+ds_file = path.split("/").last
+
 if File.exists?(path)
   puts "#{ds_file} exists"
 else
@@ -29,7 +32,7 @@ pc_type = ARGV[1] # e.g. electronic,cpsa or nil to disable
 algo = ARGV[2]    # e.g. local_svm_regression, local_mlr_prop
 r_seed = ARGV[3]  # 1, 2, ..., 10
 
-ds = YAML::load_file("../data/datasets.yaml")
+ds = YAML::load_file("#{path}")
 ds_uri = ds[ds_name]["dataset"]
 pc_ds_uri = ds[ds_name][pc_type]
 
diff --git a/cv/scripts/wrapper_pc_cv_LOAEL.sh b/cv/scripts/wrapper_pc_cv_LOAEL.sh
index 4d74596..e4f5685 100755
--- a/cv/scripts/wrapper_pc_cv_LOAEL.sh
+++ b/cv/scripts/wrapper_pc_cv_LOAEL.sh
@@ -3,14 +3,15 @@
 # Reads factors_config, ../data/datasets.yaml and performs cv's
 # Andreas Maunz, David Vorgrimmler,  2012
 
-if [ $# -lt 1 ]; then
-  echo "Usage: $0 factors"
+if [ $# -lt 2 ]; then
+  echo "Usage: $0 factors path/to/dataset.yaml"
   exit
 fi
 
-PWD=`pwd`
-echo $PWD
-if [ ! -f $PWD/../data/datasets.yaml ] 
+#PWD=`pwd`
+#echo $PWD
+#if [ ! -f $PWD/../data/datasets.yaml ] 
+if [ ! -f $2 ] 
 then
   echo "datasets.yaml does not exist."
   exit
@@ -36,8 +37,8 @@ cat $FACTORS | while read factor; do
     do
       factor="$factor $r_seed"
       echo "${THIS_DATE}: $factor" >> $LOGFILE>&1
-      echo "ruby $CV $factor" >> $LOGFILE 2>&1
-      ruby $CV $factor >> $LOGFILE 2>&1
+      echo "ruby $CV $factor $2" >> $LOGFILE 2>&1
+      ruby $CV $factor $2>> $LOGFILE 2>&1
       echo >> $LOGFILE 2>&1
     done
   fi
diff --git a/cv/scripts/wrapper_pc_cv_LOAEL_loo.sh b/cv/scripts/wrapper_pc_cv_LOAEL_loo.sh
index be3e6df..f75f385 100755
--- a/cv/scripts/wrapper_pc_cv_LOAEL_loo.sh
+++ b/cv/scripts/wrapper_pc_cv_LOAEL_loo.sh
@@ -3,14 +3,15 @@
 # Reads factors_config, ../data/datasets.yaml and performs cv's
 # Andreas Maunz, David Vorgrimmler,  2012
 
-if [ $# -lt 1 ]; then
-  echo "Usage: $0 factors"
+if [ $# -lt 2 ]; then
+  echo "Usage: $0 factors path/to/dataset.yaml"
   exit
 fi
 
-PWD=`pwd`
-echo $PWD
-if [ ! -f $PWD/../data/datasets.yaml ] 
+#PWD=`pwd`
+#echo $PWD
+#if [ ! -f $PWD/../data/datasets.yaml ] 
+if [ ! -f $2 ] 
 then
   echo "datasets.yaml does not exist."
   exit
@@ -36,8 +37,8 @@ cat $FACTORS | while read factor; do
     do
       factor="$factor $r_seed"
       echo "${THIS_DATE}: $factor" >> $LOGFILE>&1
-      echo "ruby $CV $factor" >> $LOGFILE 2>&1
-      ruby $CV $factor >> $LOGFILE 2>&1
+      echo "ruby $CV $factor $2" >> $LOGFILE 2>&1
+      ruby $CV $factor $2>> $LOGFILE 2>&1
       echo >> $LOGFILE 2>&1
     done
   fi