statsample 0.13.1 → 0.14.0

data/lib/statsample/factor/pca.rb

@@ -29,27 +29,42 @@ module Factor
   # * Smith, L. (2002). A tutorial on Principal Component Analysis. Available on http://courses.eas.ualberta.ca/eas570/pca_tutorial.pdf 
   # 
   class PCA
+    include Summarizable
     # Name of analysis
     attr_accessor :name
+
     # Number of factors. Set by default to the number of factors
     # with eigen values > 1
     attr_accessor :m
     # Use GSL if available
     attr_accessor :use_gsl
-    include Summarizable
-    
+    # Add to the summary a rotation report
+    attr_accessor :summary_rotation
+    # Add to the summary a parallel analysis report
+    attr_accessor :summary_parallel_analysis
+    # Type of rotation. By default, Statsample::Factor::Rotation::Varimax
+    attr_accessor :rotation_type
+
     def initialize(matrix, opts=Hash.new)
-	@use_gsl=nil
+      @use_gsl=nil
       @name=_("Principal Component Analysis")
       @matrix=matrix
       @n_variables=@matrix.column_size
       @m=nil
+      
+      @rotation_type=Statsample::Factor::Varimax
+      
       opts.each{|k,v|
         self.send("#{k}=",v) if self.respond_to? k
       }
       if @use_gsl.nil?
         @use_gsl=Statsample.has_gsl?
       end
+      if @matrix.respond_to? :fields
+        @variables_names=@matrix.fields
+      else
+        @variables_names=@n_variables.times.map {|i| "V#{i+1}"}
+      end
       calculate_eigenpairs
       if @m.nil?
         # Set number of factors with eigenvalues > 1
@@ -57,6 +72,9 @@ module Factor
       end
 
     end
+    def rotation
+      @rotation_type.new(component_matrix)
+    end
     def create_centered_ds
       h={}
       @original_ds.factors.each {|f|
@@ -93,7 +111,13 @@ module Factor
         gammas.push(Math::sqrt(@eigenpairs[i][0]))
       }
       gamma_m=::Matrix.diagonal(*gammas)
-      (omega_m*(gamma_m)).to_matrix
+      cm=(omega_m*(gamma_m)).to_matrix
+      
+      cm.extend CovariateMatrix
+      cm.name=_("Component matrix")
+      cm.fields_x = @variables_names
+      cm.fields_y = m.times.map {|i| "component_#{i+1}"}
+      cm
     end
     # Communalities for all variables given m factors
     def communalities(m=nil)
@@ -122,12 +146,7 @@ module Factor
     end
   
     def calculate_eigenpairs_ruby
-      eigval, eigvec= @matrix.eigenvaluesJacobi, @matrix.cJacobiV
-      @eigenpairs={}
-      eigval.to_a.each_index {|i|
-        @eigenpairs[eigval[i]]=eigvec.column(i)
-      }
-      @eigenpairs=@eigenpairs.sort.reverse
+      @eigenpairs = @matrix.eigenpairs
     end
     def calculate_eigenpairs_gsl
       eigval, eigvec= GSL::Eigen.symmv(@matrix.to_gsl)
@@ -140,25 +159,24 @@ module Factor
     
     def report_building(builder) # :nodoc:
       builder.section(:name=>@name) do |generator|
-      generator.text _("Number of factors: %d") % m
-      generator.table(:name=>_("Communalities"), :header=>[_("Variable"),_("Initial"),_("Extraction")]) do |t|
-        communalities(m).each_with_index {|com,i|
-          t.row([i, 1.0, sprintf("%0.3f", com)])
-        }
-      end      
-      generator.table(:name=>_("Eigenvalues"), :header=>[_("Variable"),_("Value")]) do |t|
-        eigenvalues.each_with_index {|eigenvalue,i|
-          t.row([i, sprintf("%0.3f",eigenvalue)])
-        }
-      end
-      
-      generator.table(:name=>_("Component Matrix"), :header=>[_("Variable")]+m.times.collect {|c| c+1}) do |t|
-        i=0
-        component_matrix(m).to_a.each do |row|
-          t.row([i]+row.collect {|c| sprintf("%0.3f",c)})
-          i+=1
+        generator.text _("Number of factors: %d") % m
+        generator.table(:name=>_("Communalities"), :header=>[_("Variable"),_("Initial"),_("Extraction")]) do |t|
+          communalities(m).each_with_index {|com, i|
+            t.row([@variables_names[i], 1.0, sprintf("%0.3f", com)])
+          }
+        end
+        
+        generator.table(:name=>_("Total Variance Explained"), :header=>[_("Component"), _("E.Total"), _("%"), _("Cum. %")]) do |t|
+          ac_eigen=0
+          eigenvalues.each_with_index {|eigenvalue,i|
+            ac_eigen+=eigenvalue
+            t.row([_("Component %d") % (i+1), sprintf("%0.3f",eigenvalue), sprintf("%0.3f%%", eigenvalue*100.quo(@n_variables)), sprintf("%0.3f",ac_eigen*100.quo(@n_variables))])
+          }
+        end
+        generator.parse_element(component_matrix(m))
+        if (summary_rotation)
+          generator.parse_element(rotation)
         end
-      end
       end
     end
     private :calculate_eigenpairs, :create_centered_ds

data/lib/statsample/factor/principalaxis.rb

@@ -29,31 +29,39 @@ module Factor
   class PrincipalAxis
     include DirtyMemoize
     include Summarizable
-    # Minimum difference between succesive iterations on sum of communalities
-    DELTA=1e-3
-    # Maximum number of iterations
-    MAX_ITERATIONS=50
-    # Number of factors. Set by default to the number of factors
-    # with eigen values > 1 on PCA over data
-    attr_accessor :m
-    
     # Name of analysis
     attr_accessor :name
+
+    # Number of factors. Set by default to the number of factors
+    # with eigenvalues > 1 (Kaiser criterion).
+    # 
+    # _Warning:_ Kaiser criterion overfactors! Give yourself some time
+    # and use Horn's Parallel Analysis.
+    #
+    attr_accessor :m
     
     # Number of iterations required to converge
     attr_reader :iterations
+    
     # Initial eigenvalues 
     attr_reader :initial_eigenvalues
-    # Tolerance for iteratios.
+    
+    # Tolerance for iterations
     attr_accessor :epsilon
+    
     # Use SMC(squared multiple correlations) as diagonal. If false, use 1
     attr_accessor :smc
+    
     # Maximum number of iterations
     attr_accessor :max_iterations
+    
     # Eigenvalues of factor analysis
     attr_reader :eigenvalues
     
-    
+    # Minimum difference between succesive iterations on sum of communalities
+    DELTA=1e-3
+    # Maximum number of iterations
+    MAX_ITERATIONS=25
     
     def initialize(matrix, opts=Hash.new)
       @matrix=matrix
@@ -62,7 +70,7 @@ module Factor
       else
         @fields=@matrix.row_size.times.map {|i| _("Variable %d") % (i+1)}
       end
-      
+      @n_variables=@matrix.row_size
       @name=""
       @m=nil
       @initial_eigenvalues=nil
@@ -74,7 +82,11 @@ module Factor
       opts.each{|k,v|
         self.send("#{k}=",v) if self.respond_to? k
       }
-      
+      if @matrix.respond_to? :fields
+        @variables_names=@matrix.fields
+      else
+        @variables_names=@n_variables.times.map {|i| "V#{i+1}"}
+      end
       if @m.nil?
         pca=PCA.new(::Matrix.rows(@matrix.to_a))
         @m=pca.m
@@ -119,7 +131,7 @@ module Factor
         pca=PCA.new(::Matrix.rows(work_matrix))
         @communalities=pca.communalities(m)
         @eigenvalues=pca.eigenvalues
-        com_sum=@communalities.inject(0) {|ac,v| ac+v}
+        com_sum = @communalities.inject(0) {|ac,v| ac+v}
         jump=true
         
         break if (com_sum-prev_sum).abs<@delta
@@ -131,6 +143,11 @@ module Factor
         
       end
       @component_matrix=pca.component_matrix(m)
+      @component_matrix.extend CovariateMatrix
+      @component_matrix.name=_("Factor Matrix")
+      @component_matrix.fields_x = @variables_names
+      @component_matrix.fields_y = m.times.map {|i| "factor_#{i+1}"}
+      
     end
     alias :compute :iterate 
     
@@ -182,18 +199,33 @@ module Factor
             t.row([@fields[i], sprintf("%0.4f", initial_communalities[i]), sprintf("%0.3f", com)])
           }
         end
-        s.table(:name=>_("Eigenvalues"), :header=>[_("Variable"),_("Value")]) do |t|
+        s.table(:name=>_("Total Variance"), :header=>[_("Factor"), _("I.E.Total"), _("I.E. %"), _("I.E.Cum. %"),
+        _("S.L.Total"), _("S.L. %"), _("S.L.Cum. %")
+          ]) do |t|
+        ac_eigen,ac_i_eigen=0,0
           @initial_eigenvalues.each_with_index {|eigenvalue,i|
-            t.row([@fields[i], sprintf("%0.3f",eigenvalue)])
+            ac_i_eigen+=eigenvalue
+            ac_eigen+=@eigenvalues[i]
+            new_row=[
+            _("Factor %d") % (i+1), 
+            sprintf("%0.3f",eigenvalue),
+            sprintf("%0.3f%%", eigenvalue*100.quo(@n_variables)),
+            sprintf("%0.3f",ac_i_eigen*100.quo(@n_variables))
+            ]
+            if i<@m
+              new_row.concat [
+                sprintf("%0.3f", @eigenvalues[i]),
+                sprintf("%0.3f%%", @eigenvalues[i]*100.quo(@n_variables)),
+                sprintf("%0.3f",ac_eigen*100.quo(@n_variables))              
+              ]
+            else
+              new_row.concat ["","",""]
+            end
+            
+            t.row new_row
           }
         end
-        s.table(:name=>_("Component Matrix"), :header=>["Variable"]+m.times.collect {|c| c+1}) do |t|
-          i=0
-          component_matrix(m).to_a.each do |row|
-            t.row([@fields[i]]+row.collect {|c| sprintf("%0.3f",c)})
-            i+=1
-          end
-        end
+        s.parse_element(component_matrix)
       end
     end
     

data/lib/statsample/factor/rotation.rb

@@ -1,10 +1,12 @@
 module Statsample
 module Factor
-  # Base class for rotate matrixes
-  # References: 
+  # Base class for component matrix rotation.
+  #
+  # == References: 
   # * SPSS Manual
-  # * Johnny Lin code for IDL: http://www.johnny-lin.com/idl_code/varimax_k58.pro
-  # Use Varimax, Equimax or Quartimax for desired type of rotation
+  # * Johnny Lin code for IDL: [http://www.johnny-lin.com/idl_code/varimax_k58.pro]
+  # 
+  # Use subclasses Varimax, Equimax or Quartimax for desired type of rotation
   #   Use:
   #   a = Matrix[ [ 0.4320,  0.8129,  0.3872] 
   #     , [ 0.7950, -0.5416,  0.2565]  
@@ -18,6 +20,7 @@ module Factor
   class Rotation
     EPSILON=1e-15
     MAX_ITERATIONS=25
+    include Summarizable
     include DirtyMemoize
     attr_reader :iterations, :rotated, :component_transformation_matrix, :h2
     # Maximum number of iterations    
@@ -29,6 +32,7 @@ module Factor
     dirty_memoize :iterations, :rotated, :component_transformation_matrix, :h2
     
     def initialize(matrix, opts=Hash.new)
+      @name=_("%s rotation") % rotation_name
       @matrix=matrix
       @n=@matrix.row_size # Variables, p on original
       @m=@matrix.column_size # Factors, r on original
@@ -41,6 +45,12 @@ module Factor
         self.send("#{k}=",v) if self.respond_to? k
       }
     end
+    def report_building(g)
+      g.section(:name=>@name) do |s|
+        s.parse_element(rotated)
+        s.parse_element(component_transformation_matrix)
+      end
+    end
     alias_method :communalities, :h2
     alias_method :rotated_component_matrix, :rotated
     def compute
@@ -111,7 +121,33 @@ module Factor
         end #i
       end # while
       @rotated=h*bh
+      @rotated.extend CovariateMatrix
+      @rotated.name=_("Rotated Component matrix")
+      
+      if @matrix.respond_to? :fields_x
+        @rotated.fields_x = @matrix.fields_x
+      else
+        @rotated.fields_x = @n.times.map {|i| "var_#{i+1}"}
+      end
+      if @matrix.respond_to? :fields_y
+        @rotated.fields_y = @matrix.fields_y
+      else
+        @rotated.fields_y = @m.times.map {|i| "var_#{i+1}"}
+      end
+      
+      
+      
       @component_transformation_matrix=t
+      @component_transformation_matrix.extend CovariateMatrix
+      @component_transformation_matrix.name=_("Component transformation matrix")
+      
+      if @matrix.respond_to? :fields_y
+        @component_transformation_matrix.fields = @matrix.fields_y
+        
+      else
+        @component_transformation_matrix.fields = @m.times.map {|i| "var_#{i+1}"}
+      end
+      
       @rotated
     end
 
@@ -123,6 +159,9 @@ module Factor
     def y(a,b,c,d)
       c-((a**2-b**2) / @n.to_f)
     end
+    def rotation_name
+      "Varimax"
+    end
   end
   class Equimax < Rotation
     def x(a,b,c,d)
@@ -131,6 +170,10 @@ module Factor
     def y(a,b,c,d)
       c-@m*((a**2-b**2) / (2*@n.to_f))
     end
+    def rotation_name
+      "Equimax"
+    end
+
   end
   class Quartimax < Rotation
     def x(a,b,c,d)
@@ -139,6 +182,10 @@ module Factor
     def y(a,b,c,d)
       c
     end
+    def rotation_name
+      "Quartimax"
+    end
+    
   end
 end
 end

data/lib/statsample/matrix.rb

@@ -57,20 +57,20 @@ module Statsample
     end
     def correlation
       if(type==:covariance)
-      matrix=Matrix.rows(row_size.times.collect { |i|
-        column_size.times.collect { |j|
-          if i==j
-            1.0
-          else
-            self[i,j].quo(Math::sqrt(self[i,i])*Math::sqrt(self[j,j]))
-          end
-        }
-      })
-      matrix.extend CovariateMatrix 
-      matrix.fields_x=fields_x
-      matrix.fields_y=fields_y
-      matrix.type=:correlation
-      matrix
+        matrix=Matrix.rows(row_size.times.collect { |i|
+          column_size.times.collect { |j|
+            if i==j
+              1.0
+            else
+              self[i,j].quo(Math::sqrt(self[i,i])*Math::sqrt(self[j,j]))
+            end
+          }
+        })
+        matrix.extend CovariateMatrix 
+        matrix.fields_x=fields_x
+        matrix.fields_y=fields_y
+        matrix.type=:correlation
+        matrix
       else
         self
       end

data/lib/statsample/reliability.rb

@@ -30,6 +30,7 @@ module Statsample
       # Get Cronbach's alpha from a covariance matrix
       def cronbach_alpha_from_covariance_matrix(cov)
         n=cov.row_size
+        raise "covariance matrix should have at least 2 variables" if n < 2
         s2=n.times.inject(0) {|ac,i| ac+cov[i,i]}
         (n.quo(n-1))*(1-(s2.quo(cov.total_sum)))
       end

data/lib/statsample/reliability/multiscaleanalysis.rb

@@ -32,10 +32,14 @@ module Statsample
       attr_accessor :summary_pca
       # Add Principal Axis to summary
       attr_accessor :summary_principal_axis
+      # Add Parallel Analysis to summary
+      attr_accessor :summary_parallel_analysis
       # Options for Factor::PCA object
       attr_accessor :pca_options
       # Options for Factor::PrincipalAxis 
       attr_accessor :principal_axis_options
+      # Options for Parallel Analysis
+      attr_accessor :parallel_analysis_options
       # Generates a new MultiScaleAnalysis
       # Opts could be any accessor of the class 
       # * :name, 
@@ -50,12 +54,15 @@ module Statsample
       # 
       def initialize(opts=Hash.new, &block)
         @scales=Hash.new
+        @scales_keys=Array.new
         opts_default={  :name=>_("Multiple Scale analysis"),
                         :summary_correlation_matrix=>false,
                         :summary_pca=>false,
                         :summary_principal_axis=>false,
+                        :summary_parallel_analysis=>false,
                         :pca_options=>Hash.new,
-                        :principal_axis_options=>Hash.new
+                        :principal_axis_options=>Hash.new,
+                        :parallel_analysis_options=>Hash.new
         }
         @opts=opts_default.merge(opts)
         @opts.each{|k,v|
@@ -72,38 +79,49 @@ module Statsample
       # 
       # If second parameters is empty, returns the ScaleAnalysis
       # <tt>code</tt>.
-      def scale(code,ds=nil, opts=nil)
+      def scale(code, ds=nil, opts=nil)
         if ds.nil?
           @scales[code]
         else
           opts={:name=>_("Scale %s") % code} if opts.nil?
+          @scales_keys.push(code)
           @scales[code]=ScaleAnalysis.new(ds, opts)
         end
       end
       # Delete ScaleAnalysis named <tt>code</tt>
       def delete_scale(code)
+        @scales_keys.delete code
         @scales.delete code
       end
       # Retrieves a Principal Component Analysis (Factor::PCA)
       # using all scales, using <tt>opts</tt> a options.
       def pca(opts=nil)
         opts||=pca_options        
-        Statsample::Factor::PCA.new(correlation_matrix,opts)
+        Statsample::Factor::PCA.new(correlation_matrix, opts)
       end
       # Retrieves a PrincipalAxis Analysis (Factor::PrincipalAxis)
       # using all scales, using <tt>opts</tt> a options.
       def principal_axis_analysis(opts=nil)
         opts||=principal_axis_options
-        Statsample::Factor::PrincipalAxis.new(correlation_matrix,opts)
+        Statsample::Factor::PrincipalAxis.new(correlation_matrix, opts)
+      end
+      def dataset_from_scales
+        ds=Dataset.new(@scales_keys)
+        @scales.each_pair do |code,scale|
+          ds[code.to_s]=scale.ds.vector_sum
+          ds[code.to_s].name=scale.name
+        end
+        ds.update_valid_data
+        ds
+      end
+      def parallel_analysis(opts=nil)
+        opts||=parallel_analysis_options
+        Statsample::Factor::ParallelAnalysis.new(dataset_from_scales, opts)
       end
       # Retrieves a Correlation Matrix between scales.
       # 
       def correlation_matrix
-        vectors=Hash.new
-        @scales.each_pair do |code,scale|
-          vectors[code.to_s]=scale.ds.vector_sum
-        end
-        Statsample::Bivariate.correlation_matrix(vectors.to_dataset)
+        Statsample::Bivariate.correlation_matrix(dataset_from_scales)
       end
       def report_building(b) # :nodoc:
         b.section(:name=>name) do |s|
@@ -126,7 +144,14 @@ module Statsample
             s.section(:name=>_("Principal Axis for %s") % name) do |s2|
               s2.parse_element(principal_axis_analysis)
             end
-          end          
+          end
+          
+          if summary_parallel_analysis
+            s.section(:name=>_("Parallel Analysis for %s") % name) do |s2|
+              s2.parse_element(parallel_analysis)
+            end
+          end 
+          
         end
       end
     end