statsample 0.6.3 → 0.6.4

data/History.txt

@@ -1,6 +1,12 @@
+=== 0.6.4 / 2010-02-19
+* Dominance Analysis and Dominance Analysis Bootstrap allows multivariate dependent analysis. 
+* Test suite for Dominance Analysis, using Azen and Budescu papers as references
+* X^2 for polychoric correlation
+
 === 0.6.3 / 2010-02-15
 * Statsample::Bivariate::Polychoric have joint estimation.
 * Some extra documentation and bug fixs
+
 === 0.6.2 / 2010-02-11
 * New Statsample::Bivariate::Polychoric. For implement: X2 and G2
 * New matrix.rb, for faster development of Contingence Tables and Correlation Matrix

data/Manifest.txt

@@ -13,6 +13,7 @@ demo/correlation_matrix.rb
 demo/dominance_analysis_bootstrap.rb
 demo/dominanceanalysis.rb
 demo/multiple_regression.rb
+demo/multivariate_correlation.rb
 demo/polychoric.rb
 demo/tetrachoric.rb
 lib/distribution.rb
@@ -63,6 +64,7 @@ lib/statsample/regression/multiple.rb
 lib/statsample/regression/multiple/alglibengine.rb
 lib/statsample/regression/multiple/baseengine.rb
 lib/statsample/regression/multiple/gslengine.rb
+lib/statsample/regression/multiple/matrixengine.rb
 lib/statsample/regression/multiple/rubyengine.rb
 lib/statsample/regression/simple.rb
 lib/statsample/reliability.rb
@@ -83,11 +85,13 @@ test/test_csv.csv
 test/test_csv.rb
 test/test_dataset.rb
 test/test_distribution.rb
+test/test_dominance_analysis.rb
 test/test_factor.rb
 test/test_ggobi.rb
 test/test_gsl.rb
 test/test_histogram.rb
 test/test_logit.rb
+test/test_matrix.rb
 test/test_mle.rb
 test/test_multiset.rb
 test/test_permutation.rb

data/README.txt

@@ -11,16 +11,16 @@ A suite for basic and advanced statistics. Includes:
 * Correlations: Pearson (r), Rho, Tetrachoric, Polychoric
 * Regression: Simple, Multiple, Probit and Logit
 * Factorial Analysis: Extraction (PCA and Principal Axis) and Rotation (Varimax and relatives)
-* Dominance Analysis (Azen & Budescu)
+* Dominance Analysis, with multivariate dependent and bootstrap (Azen & Budescu)
 * Sample calculation related formulas
 
 == DETAILED FEATURES:
 
 * Factorial Analysis. Principal Component Analysis and Principal Axis extraction, with orthogonal rotations (Varimax, Equimax, Quartimax)
-* Multiple Regression. Listwise analysis optimized with use of Alglib library. Pairwise analysis is executed on pure ruby and reports same values as SPSS
+* Multiple Regression. Listwise analysis optimized with use of Alglib library. Pairwise analysis is executed on pure ruby with matrixes and reports same values as SPSS
 * Module Bivariate provides covariance and pearson, spearman, point biserial, tau a, tau b, gamma, tetrachoric and polychoric correlation correlations. Include methods to create correlation (pearson and tetrachoric) and covariance matrices
 * Regression module provides linear regression methods
-* Dominance Analysis. Based on Budescu and Azen papers, <strong>DominanceAnalysis</strong> class can report dominance analysis for a sample and <strong>DominanceAnalysisBootstrap</strong> can execute bootstrap analysis to determine dominance stability, as recomended by  Azen & Budescu (2003) link[http://psycnet.apa.org/journals/met/8/2/129/]. 
+* Dominance Analysis. Based on Budescu and Azen papers, <strong>DominanceAnalysis</strong> class can report dominance analysis for a sample, using uni or multivariate dependent variables and <strong>DominanceAnalysisBootstrap</strong> can execute bootstrap analysis to determine dominance stability, as recomended by  Azen & Budescu (2003) link[http://psycnet.apa.org/journals/met/8/2/129/]. 
 * Classes for Vector, Datasets (set of Vectors) and Multisets (multiple datasets with same fields and type of vectors), and multiple methods to manipulate them
 * Module Codification, to help to codify open questions
 * Converters to and from database and csv files, and to output Mx and GGobi files
@@ -66,7 +66,7 @@ A suite for basic and advanced statistics. Includes:
 Optional: 
 
 * Plotting: gnuplot and rbgnuplot, SVG::Graph
-* Factorial analysis and polychorical correlation: gsl and rb-gsl (http://rb-gsl.rubyforge.org/)
+* Factorial analysis and polychorical correlation: gsl library and rb-gsl (http://rb-gsl.rubyforge.org/). You should install it using <tt>gem install gsl</tt>
 
 == DOWNLOAD
 * Gems and bugs report: http://rubyforge.org/projects/ruby-statsample/
@@ -78,7 +78,7 @@ Optional:
 
 For optimization on *nix env
 
-  sudo gem install ruby-statsample-optimization
+  sudo gem install gsl ruby-statsample-optimization
 
 Available setup.rb file
 

data/demo/dominance_analysis_bootstrap.rb

@@ -10,7 +10,13 @@ d=100.times.collect {rand}.to_scale
 
 ds={'a'=>a,'b'=>b,'c'=>c,'d'=>d}.to_dataset
 
-ds['y']=ds.collect{|row| row['a']*5+row['b']*2+row['c']*2+row['d']*2+10*rand()}
-dab=Statsample::DominanceAnalysis::Bootstrap.new(ds, 'y')
-dab.bootstrap(100,nil,true)
+ds['y1']=ds.collect{|row| row['a']*5+row['b']*2+row['c']*2+row['d']*2+10*rand()}
+ds['y2']=ds.collect{|row| row['a']*10+rand()}
+
+dab=Statsample::DominanceAnalysis::Bootstrap.new(ds, ['y1','y2'], :debug=>true)
+dab.bootstrap(100,nil)
+puts dab.summary
+ds2=ds['a'..'y1']
+dab=Statsample::DominanceAnalysis::Bootstrap.new(ds2, 'y1', :debug=>true)
+dab.bootstrap(100,nil)
 puts dab.summary

data/demo/dominanceanalysis.rb

@@ -2,10 +2,26 @@
 $:.unshift(File.dirname(__FILE__)+'/../lib/')
 
 require 'statsample'
-a=1000.times.collect {rand}.to_scale
-b=1000.times.collect {rand}.to_scale
-c=1000.times.collect {rand}.to_scale
-ds={'a'=>a,'b'=>b,'c'=>c}.to_dataset
-ds['y']=ds.collect{|row| row['a']*5+row['b']*3+row['c']*2+rand()}
-da=Statsample::DominanceAnalysis.new(ds,'y')
-puts da.summary
+sample=200
+a=sample.times.collect {rand}.to_scale
+b=sample.times.collect {rand}.to_scale
+c=sample.times.collect {rand}.to_scale
+d=sample.times.collect {rand}.to_scale
+
+ds={'a'=>a,'b'=>b,'c'=>c,'d'=>d}.to_dataset
+ds['y']=ds.collect{|row| row['a']*5+row['b']*3+row['c']*2+row['d']+rand()}
+rb=ReportBuilder.new("Dominance Analysis")
+
+cm=Statsample::Bivariate.correlation_matrix(ds)
+rb.add(cm)
+lr=Statsample::Regression::Multiple::RubyEngine.new(ds,'y')
+rb.add(lr)
+
+#da=Statsample::DominanceAnalysis.new(ds,'y')
+#rb.add(da)
+
+da=Statsample::DominanceAnalysis.new(ds,'y',:name=>"Dominance Analysis using group of predictors", :predictors=>['a', 'b', %w{c d}])
+rb.add(da)
+
+
+puts rb.to_text

data/demo/multivariate_correlation.rb

@@ -0,0 +1,26 @@
+#!/usr/bin/ruby
+$:.unshift(File.dirname(__FILE__)+'/../lib/')
+
+require 'statsample'
+require 'mathn'
+
+
+
+  complete=Matrix[
+  [1,0.53,0.62,0.19,-0.09,0.08,0.02,-0.12,0.08],
+  [0.53,1,0.61,0.23,0.1,0.18,0.02,-0.1,0.15],
+  [0.62,0.61,1,0.03,0.1,0.12,0.03,-0.06,0.12],
+  [0.19,0.23,0.03,1,-0.02,0.02,0,-0.02,-0.02],
+  [-0.09,0.1,0.1,-0.02,1,0.05,0.06,0.18,0.02],
+  [0.08,0.18,0.12,0.02,0.05,1,0.22,-0.07,0.36],
+  [0.02,0.02,0.03,0,0.06,0.22,1,-0.01,-0.05],
+  [-0.12,-0.1,-0.06,-0.02,0.18,-0.07,-0.01,1,-0.03],
+  [0.08,0.15,0.12,-0.02,0.02,0.36,-0.05,-0.03,1]]
+  complete.extend Statsample::CovariateMatrix
+  complete.fields=%w{adhd cd odd sex age monly mwork mage poverty}
+  
+  lr=Statsample::Regression::Multiple::MultipleDependent.new(complete, %w{adhd cd odd})
+  
+  puts "R^2_yx #{lr.r2yx}"
+  puts "P^2_yx #{lr.p2yx}"
+

data/lib/statsample.rb

@@ -109,7 +109,7 @@ end
 # * Dataset: An union of vectors.
 #
 module Statsample
-  VERSION = '0.6.3'
+  VERSION = '0.6.4'
   SPLIT_TOKEN = ","
   autoload(:Database, 'statsample/converters')
   autoload(:Anova, 'statsample/anova')

data/lib/statsample/bivariate.rb

@@ -120,13 +120,18 @@ module Statsample
       # Order of rows and columns depends on Dataset#fields order
       
       def covariance_matrix(ds)
-        ds.collect_matrix do |row,col|
+        matrix=ds.collect_matrix do |row,col|
           if (ds[row].type!=:scale or ds[col].type!=:scale)
             nil
+          elsif row==col
+            ds[row].variance
           else
-            covariance(ds[row],ds[col])
+            covariance(ds[row], ds[col])
           end
         end
+        matrix.extend CovariateMatrix
+        matrix.fields=ds.fields
+        matrix
       end
       
       # Correlation matrix.
@@ -142,8 +147,8 @@ module Statsample
             pearson(ds[row],ds[col])
           end
         end
-        cm.extend(Statsample::CorrelationMatrix)
-        cm.labels=ds.fields
+        cm.extend(Statsample::CovariateMatrix)
+        cm.fields=ds.fields
         cm
       end
       
@@ -282,6 +287,21 @@ module Statsample
         }
         sum-((v1a.sum*v2a.sum) / v1a.size.to_f)
       end
+      
+      # Report the minimum number of cases valid of a covariate matrix
+      # based on a dataset
+      def min_n_valid(ds)
+        min=ds.cases
+        m=n_valid_matrix(ds)
+        for x in 0...m.row_size
+          for y in 0...m.column_size
+            min=m[x,y] if m[x,y] < min
+          end
+        end
+        min
+      end
+      
+      
     end
   end
 end

data/lib/statsample/bivariate/polychoric.rb

@@ -48,21 +48,21 @@ module Statsample
       attr_accessor :max_iterations
       # Debug algorithm (See iterations, for example)
       attr_accessor :debug
-      # Minimizer type. Default GSL::Min::FMinimizer::BRENT
+      # Minimizer type for two step. Default "brent"
       # See http://rb-gsl.rubyforge.org/min.html for reference.  
       attr_accessor :minimizer_type_two_step
       
-      # Minimizer type. Default GSL::Min::FMinimizer::BRENT
+      # Minimizer type for joint estimate. Default "nmsimplex"
       # See http://rb-gsl.rubyforge.org/min.html for reference.  
       attr_accessor :minimizer_type_joint
       
       
       # Method of calculation of polychoric series. 
       # 
-      # :two_step:: two-step ML, based on code by Gegenfurtner(1992)
+      # :two_step:: two-step ML, based on code by Gegenfurtner(1992).
       # :polychoric_series:: polychoric series estimate, using 
-      #                      algorithm AS87 by Martinson and Hamdan (1975)
-      # :joint:              one-step ML, based on R package 'polycor' 
+      #                      algorithm AS87 by Martinson and Hamdan (1975).
+      # :joint::             one-step ML, based on R package 'polycor'
       #                      by J.Fox.
       attr_accessor :method
       # Absolute error for iteration.
@@ -73,7 +73,9 @@ module Statsample
       
       # Log of algorithm
       attr_reader :log
-      attr_reader :loglike
+      
+      
+      attr_reader :loglike_model
       
       METHOD=:two_step
       MAX_ITERATIONS=300
@@ -162,16 +164,15 @@ module Statsample
       
       def loglike_data
         loglike=0
-        @nr.times { |i|
-          @nc.times { |j|
+        @nr.times do |i|
+          @nc.times do |j|
             res=@matrix[i,j].quo(@total)
             if (res==0)
-           #    puts "Correccion"
-            res=1e-16
-          end
+              res=1e-16
+            end
           loglike+= @matrix[i,j]  * Math::log(res )
-          }
-        }
+          end
+        end
         loglike
       end
       def chi_square
@@ -346,7 +347,7 @@ module Statsample
           end
           message+=sprintf("f() = %7.3f size = %.3f\n", minimizer.fval, minimizer.size)+"\n";
         end while status == GSL::CONTINUE and iter < @max_iterations
-        @iteration=@iter
+        @iteration=iter
         @log+=message
         puts message if @debug
         @r=minimizer.x[0]

data/lib/statsample/converters.rb

@@ -155,6 +155,28 @@ raise "Should'nt be empty headers: [#{row.to_a.join(",")}]" if row.to_a.find_all
         }
         book.write(filename)
       end
+      # This should be fixed.
+      # If we have a Formula, should be resolver first
+
+      def preprocess_row(row, dates)
+        i=-1
+        row.collect!{|c|
+          i+=1
+          if c.is_a? Spreadsheet::Formula
+            if(c.value.is_a? Spreadsheet::Excel::Error)
+              nil
+            else
+              c.value
+            end
+          elsif dates.include? i and !c.nil? and c.is_a? Numeric
+              row.date(i)
+          else
+              c
+          end
+        }
+      end
+      private :process_row
+      
       # Returns a dataset based on a xls file
       # USE:
       #     ds = Statsample::Excel.read("test.xls")
@@ -177,27 +199,9 @@ raise "Should'nt be empty headers: [#{row.to_a.join(",")}]" if row.to_a.find_all
               end
             }
             line_number+=1
-            if(line_number<=ignore_lines)
-            #puts "Skip line #{line_number}:#{row.to_s}"
-                next
-            end
-            # This should be fixed.
-            # If we have a Formula, should be resolver first
-            i=-1
-            row.collect!{|c|
-                i+=1
-                if c.is_a? Spreadsheet::Formula
-                  if(c.value.is_a? Spreadsheet::Excel::Error)
-                    nil
-                  else
-                    c.value
-                  end
-                elsif dates.include? i and !c.nil? and c.is_a? Numeric
-                    row.date(i)
-                else
-                    c
-                end
-            }
+            next if(line_number<=ignore_lines)
+            
+            preprocess_row(row,dates)
             if first_row
               fields=extract_fields(row)
               ds=Statsample::Dataset.new(fields)
@@ -210,8 +214,8 @@ raise "Should'nt be empty headers: [#{row.to_a.join(",")}]" if row.to_a.find_all
               ds.add_case(rowa,false)
             end
           rescue => e
-              error="#{e.to_s}\nError on Line # #{line_number}:#{row.join(",")}"
-              raise
+            error="#{e.to_s}\nError on Line # #{line_number}:#{row.join(",")}"
+            raise
           end
         end
         convert_to_scale_and_date(ds, fields)

data/lib/statsample/crosstab.rb

@@ -8,7 +8,7 @@ module Statsample
     bindtextdomain("statsample")
     attr_reader :v_rows, :v_cols
     attr_accessor :row_label, :column_label, :name, :percentage_row, :percentage_column, :percentage_total
-    def initialize(v1,v2,opts=Hash.new)
+    def initialize(v1, v2, opts=Hash.new)
     raise ArgumentError, "Both arguments should be Vectors" unless v1.is_a? Statsample::Vector and v2.is_a? Statsample::Vector
     raise ArgumentError, "Vectors should be the same size" unless v1.size==v2.size
     @v_rows, @v_cols=Statsample.only_valid(v1,v2)
@@ -191,48 +191,5 @@ module Statsample
       t.add_row(t_row)
       generator.parse_element(t)
     end
-  
-  
-    
-    def to_s
-      fq=frequencies
-      rn=rows_names
-      cn=cols_names
-      total=0
-      total_cols=cols_empty_hash
-      max_row_size = rn.inject(0) {|s,x| sl=@v_rows.labeling(x).size; sl>s ? sl : s}
-      
-      max_row_size=max_row_size<6 ? 6 : max_row_size
-      
-      max_col_size = cn.inject(0) {|s,x| sl=@v_cols.labeling(x).size; sl>s ? sl : s}
-      max_col_size = frequencies.inject(max_col_size) {|s,x| x[1].to_s.size>s ? x[1].to_s.size : s}
-      
-      out=""
-      out << " " * (max_row_size+2) << "|" << cn.collect{|c| name=@v_cols.labeling(c); " "+name+(" "*(max_col_size-name.size))+" "}.join("|") << "| Total\n"
-      linea="-" * (max_row_size+2) << "|" << ("-"*(max_col_size+2) +"|")*cn.size << "-"*7 << "\n"
-      out << linea
-      rn.each{|row|
-          total_row=0;
-          name=@v_rows.labeling(row)
-          out << " " +name  << " "*(max_row_size-name.size) << " | "
-          cn.each{|col|
-              data=fq[[row,col]].to_s
-              total_row+=fq[[row,col]]
-              total+=fq[[row,col]]                    
-              total_cols[col]+=fq[[row,col]]                    
-              out << " " << data << " "*(max_col_size-data.size) << "| "
-          }
-          out << " " << total_row.to_s
-      out << "\n"
-      }
-      out << linea
-      out << " Total " << " "*(max_row_size-5) << "| "
-      cn.each{|v|
-          data=total_cols[v].to_s
-          out << " " << data << " "*(max_col_size-data.size) << "| "
-      }
-      out << " " << total.to_s
-      out
-    end
   end
 end

data/lib/statsample/dominanceanalysis.rb

@@ -4,7 +4,8 @@ module Statsample
   # for all possible subset models, to identify the relevance of one or more 
   # predictors in the prediction of criterium.
   #
-  # See Budescu(1993) and Azen & Budescu (2003) for more information.
+  #
+  # See Budescu(1993), Azen & Budescu (2003, 2006) for more information.
   #
   # Example: 
   #
@@ -53,39 +54,127 @@ module Statsample
 
   #
   # == References:
-  # * Budescu, D. V. (1993). Dominance analysis: a new approach to the problem of relative importance of predictors in multiple regression. _Psychological Bulletin, 114_, 542-551.
-  # * Azen, R. & Budescu, D.V. (2003). The dominance analysis approach for comparing predictors in multiple regression. _Psychological Methods, 8_(2), 129-148.
+  # * Budescu, D. V. (1993). Dominance analysis: a new approach to the problem of relative importance of predictors in multiple regression. <em>Psychological Bulletin, 114</em>, 542-551.
+  # * Azen, R. & Budescu, D.V. (2003). The dominance analysis approach for comparing predictors in multiple regression. <em>Psychological Methods, 8</em>(2), 129-148.
+  # * Azen, R. & Budescu, D.V. (2006). Comparing predictors in Multivariate Regression Models: An extension of Dominance Analysis. <em>Journal of Educational and Behavioral Statistics, 31</em>(2), 157-180.
   class DominanceAnalysis
     include GetText
     bindtextdomain("statsample")
-    # Class to generate the regressions. Default to Statsample::Regression::Multiple::RubyEngine
+    # Class to generate the regressions. Default to Statsample::Regression::Multiple::MatrixEngine
     attr_accessor :regression_class
     # Name of analysis
     attr_accessor :name
+    # Set to true if you want to build from dataset, not correlation matrix
+    attr_accessor :build_from_dataset
+    #  Array with independent variables. You could create subarrays, 
+    #  to test groups of predictors as blocks
+    attr_accessor  :predictors
+    # If you provide a matrix as input, you should set 
+    # the number of cases to define significance of R^2
+    attr_accessor  :cases
+    # Method of :regression_class used to measure association. 
+    # 
+    # Only necessary to change if you have multivariate dependent.
+    # * :r2yx (R^2_yx), the default option, is the  option when distinction
+    #   between independent and dependents variable is arbitrary
+    # * :p2yx is the option when the distinction between independent and dependents variables is real.
+    #   
+    
+    attr_accessor  :method_association
+    
+    
+    attr_reader :dependent
+    
+    UNIVARIATE_REGRESSION_CLASS=Statsample::Regression::Multiple::MatrixEngine
+    MULTIVARIATE_REGRESSION_CLASS=Statsample::Regression::Multiple::MultipleDependent
     
+    def self.predictor_name(variable)
+      if variable.is_a? Array
+        sprintf("(%s)", variable.join(","))
+      else
+        variable
+      end
+    end
     # Creates a new DominanceAnalysis object
-    # Params:
-    # * ds: A Dataset object
-    # * y_var: Name of dependent variable
-    # * opts: Any other attribute of the class 
-    # 
-    def initialize(ds,y_var, opts=Hash.new)
-      @y_var=y_var
-      @dy=ds[@y_var]
-      @ds=ds
-      @ds_indep=ds.dup(ds.fields-[y_var])
-      @fields=@ds_indep.fields
-      @regression_class=Statsample::Regression::Multiple::RubyEngine
-      @name=_("Dominance Analysis:  %s over %s") % [ ds.fields.join(",") , @y_var]
+    # Parameters:
+    # * input:    A Matrix or Dataset object
+    # * dependent: Name of dependent variable. Could be an array, if you want to
+    #             do an Multivariate Regression Analysis. If nil, set to all
+    #             fields on input, except criteria
+ 
+    def initialize(input, dependent, opts=Hash.new)
+      @build_from_dataset=false
+      if dependent.is_a? Array
+        @regression_class= MULTIVARIATE_REGRESSION_CLASS
+        @method_association=:r2yx
+      else
+        @regression_class= UNIVARIATE_REGRESSION_CLASS
+        @method_association=:r2
+        
+      end
       opts.each{|k,v|
         self.send("#{k}=",v) if self.respond_to? k
       }
+      @dependent=dependent
+      @dependent=[@dependent] unless @dependent.is_a? Array
+      
+      @predictors ||= input.fields-@dependent
+
+      @name=_("Dominance Analysis:  %s over %s") % [ @predictors.flatten.join(",") , @dependent.join(",")] if @name.nil?
+      
+      if input.is_a? Statsample::Dataset
+        @ds=input
+        @matrix=Statsample::Bivariate.correlation_matrix(input)
+        @cases=Statsample::Bivariate.min_n_valid(input)
+      elsif input.is_a? ::Matrix
+        @ds=nil
+        @matrix=input
+      else
+        raise ArgumentError.new("You should use a Matrix or a Dataset")
+      end
+      @models=nil
+      
+    end
+    # Compute models. 
+    def compute
       create_models
       fill_models
     end
+    def models
+      if @models.nil?
+        compute
+      end
+      @models
+    end
+    
+    def models_data
+      if @models_data.nil?
+        compute
+      end
+      @models_data
+    end
+    def create_models
+      @models=[]
+      @models_data={}
+      for i in 1..@predictors.size
+        c=Statsample::Combination.new(i,@predictors.size)
+        c.each  do |data|
+          independent=data.collect {|i1| @predictors[i1] }
+          @models.push(independent)
+          if (@build_from_dataset)
+            data=@ds.dup(independent.flatten+@dependent)
+          else
+            data=@matrix.submatrix(independent.flatten+@dependent)
+          end
+          
+          modeldata=ModelData.new(independent, data, self)
+          models_data[independent.sort {|a,b| a.to_s<=>b.to_s}]=modeldata
+        end
+      end
+    end
     def fill_models
       @models.each do |m|
-        @fields.each do |f|
+        @predictors.each do |f|
           next if m.include? f
           base_model=md(m)
           comp_model=md(m+[f])
@@ -93,6 +182,8 @@ module Statsample
         end
       end
     end
+    private :create_models, :fill_models
+    
     def dominance_for_nil_model(i,j)
       if md([i]).r2>md([j]).r2
         1
@@ -107,7 +198,7 @@ module Statsample
       dm=dominance_for_nil_model(i,j)
       return 0.5 if dm==0.5
       dominances=[dm]
-      @models_data.each do |k,m|
+      models_data.each do |k,m|
         if !m.contributions[i].nil? and !m.contributions[j].nil?
           if m.contributions[i]>m.contributions[j]
               dominances.push(1)
@@ -128,7 +219,7 @@ module Statsample
       dm=dominance_for_nil_model(i,j)
       return 0.5 if dm==0.5
       dominances=[dm]
-      for k in 1...@fields.size
+      for k in 1...@predictors.size
         a=average_k(k)
         if a[i]>a[j]
             dominances.push(1)
@@ -154,7 +245,7 @@ module Statsample
       end                 
     end
     def pairs
-      @models.find_all{|m| m.size==2}
+      models.find_all{|m| m.size==2}
     end
     def total_dominance
       pairs.inject({}){|a,pair| a[pair]=total_dominance_pairwise(pair[0], pair[1])
@@ -162,20 +253,18 @@ module Statsample
       }
     end
     def conditional_dominance
-      pairs.inject({}){|a,pair|
-      a[pair]=conditional_dominance_pairwise(pair[0], pair[1])
+      pairs.inject({}){|a,pair| a[pair]=conditional_dominance_pairwise(pair[0], pair[1])
       a
       }
     end
     def general_dominance
-      pairs.inject({}){|a,pair|
-      a[pair]=general_dominance_pairwise(pair[0], pair[1])
+      pairs.inject({}){|a,pair| a[pair]=general_dominance_pairwise(pair[0], pair[1])
       a
       }
     end
     
     def md(m)
-      @models_data[m.sort]
+      models_data[m.sort {|a,b| a.to_s<=>b.to_s}]
     end
     # Get all model of size k
     def md_k(k)
@@ -195,11 +284,11 @@ module Statsample
     end
     # Hash with average for each k size model.
     def average_k(k)
-      return nil if k==@fields.size
+      return nil if k==@predictors.size
       models=md_k(k)
-      averages=@fields.inject({}) {|a,v| a[v]=[];a}
+      averages=@predictors.inject({}) {|a,v| a[v]=[];a}
       models.each do |m|
-        @fields.each do |f|
+        @predictors.each do |f|
           averages[f].push(m.contributions[f]) unless m.contributions[f].nil?
         end
       end
@@ -207,10 +296,10 @@ module Statsample
     end
     def general_averages
       if @general_averages.nil?
-        averages=@fields.inject({}) {|a,v| a[v]=[md([v]).r2];a}
-        for k in 1...@fields.size
+        averages=@predictors.inject({}) {|a,v| a[v]=[md([v]).r2];a}
+        for k in 1...@predictors.size
           ak=average_k(k)
-          @fields.each do |f|
+          @predictors.each do |f|
             averages[f].push(ak[f])
           end
         end
@@ -218,36 +307,25 @@ module Statsample
       end
       @general_averages
     end
-    def create_models
-      @models=[]
-      @models_data={}
-      for i in 1..@fields.size
-      c=Statsample::Combination.new(i,@fields.size)
-      c.each  do |data|
-        convert=data.collect {|i1| @fields[i1] }
-        @models.push(convert)
-        ds_prev=@ds.dup(convert+[@y_var])
-        modeldata=ModelData.new(convert,ds_prev, @y_var, @fields, @regression_class)
-        @models_data[convert.sort]=modeldata
-      end
-      end
-    end
+    
     def summary
       rp=ReportBuilder.new()
       rp.add(self)
       rp.to_text
     end
     def to_reportbuilder(generator)
+      compute if @models.nil?
       anchor=generator.add_toc_entry(_("DA: ")+@name)
       generator.add_html "<div class='dominance-analysis'>#{@name}<a name='#{anchor}'></a>"
       t=ReportBuilder::Table.new(:name=>_("Dominance Analysis result"))
-      t.header=["","r2",_("sign")]+@fields
-      row=[_("Model 0"),"",""]+@fields.collect{|f|
+      
+      t.header=["","r2",_("sign")]+@predictors.collect {|c| DominanceAnalysis.predictor_name(c) }
+      row=[_("Model 0"),"",""]+@predictors.collect{|f|
         sprintf("%0.3f", md([f]).r2)
       }
       t.add_row(row)
       t.add_horizontal_line
-      for i in 1..@fields.size
+      for i in 1..@predictors.size
         mk=md_k(i)
         mk.each{|m|
           t.add_row(m.add_table_row)
@@ -256,7 +334,7 @@ module Statsample
         a=average_k(i)
         if !a.nil?
             t.add_horizontal_line
-            row=[_("k=%d Average") % i,"",""] + @fields.collect{|f|
+            row=[_("k=%d Average") % i,"",""] + @predictors.collect{|f|
                 sprintf("%0.3f",a[f])
             }
             t.add_row(row)
@@ -269,7 +347,7 @@ module Statsample
       g=general_averages
       t.add_horizontal_line
       
-      row=[_("Overall averages"),"",""]+@fields.collect{|f|
+      row=[_("Overall averages"),"",""]+@predictors.collect{|f|
                 sprintf("%0.3f",g[f])
       }
       t.add_row(row)
@@ -289,26 +367,42 @@ module Statsample
     end
     class ModelData
       attr_reader :contributions
-      def initialize(name,ds,y_var,fields,r_class)
-        @name=name
-        @fields=fields
-        @contributions=@fields.inject({}){|a,v| a[v]=nil;a}
-        r_class=Regression::Multiple::RubyEngine if r_class.nil?
-        @lr=r_class.new(ds,y_var)
+      def initialize(independent, data, da)
+        @independent=independent
+        @data=data
+        @predictors=da.predictors
+        @dependent=da.dependent
+        @cases=da.cases
+        @method=da.method_association
+        @contributions=@independent.inject({}){|a,v| a[v]=nil;a}
+        
+        r_class=da.regression_class
+        
+        if @dependent.size==1
+          @lr=r_class.new(data, @dependent[0], :cases=>@cases)
+        else
+          @lr=r_class.new(data, @dependent, :cases=>@cases)
+        end
       end
-      def add_contribution(f,v)
+      def add_contribution(f, v)
         @contributions[f]=v-r2
       end
       def r2
-        @lr.r2
+        @lr.send(@method)
+      end
+      def name
+        @independent.collect {|variable|
+          DominanceAnalysis.predictor_name(variable)
+        }.join("*")
       end
       def add_table_row
         begin
-        sign=sprintf("%0.3f", @lr.significance)
+          sign=sprintf("%0.3f", @lr.significance)
         rescue RuntimeError
-            sign="???"
+          sign="???"
         end
-        [@name.join("*"), sprintf("%0.3f",r2), sign] + @fields.collect{|k|
+      
+        [name, sprintf("%0.3f",r2), sign] + @predictors.collect{|k|
           v=@contributions[k]
           if v.nil?
               "--"
@@ -318,8 +412,8 @@ module Statsample
         }
       end
       def summary
-        out=sprintf("%s: r2=%0.3f(p=%0.2f)\n",@name.join("*"),r2,@lr.significance,@lr.sst)
-        out << @fields.collect{|k|
+        out=sprintf("%s: r2=%0.3f(p=%0.2f)\n",name, r2, @lr.significance, @lr.sst)
+        out << @predictors.collect{|k|
           v=@contributions[k]
           if v.nil?
               "--"