statsample 0.6.5 → 0.6.7

data/lib/statsample/matrix.rb

@@ -92,33 +92,43 @@ module GSL
 end
 
 module Statsample
-  # Method for variance/covariance and correlation matrices
+  # Module to add method for variance/covariance and correlation matrices
+  # == Usage
+  #  matrix=Matrix[[1,2],[2,3]]
+  #  matrix.extend CovariateMatrix
+  # 
   module CovariateMatrix
+    # Gives a nice 
     def summary
       rp=ReportBuilder.new()
       rp.add(self)
       rp.to_text
     end
-    def type=(v)
-      @type=v
-    end
+    # Get type of covariate matrix. Could be :covariance or :correlation
     def type
-      if row_size.times.find {|i| self[i,i]!=1.0}
-        :covariance
+      if row_size==column_size
+        if row_size.times.find {|i| self[i,i]!=1.0}
+          :covariance
+        else
+          :correlation
+        end
       else
-        :correlation
+        @type
       end
       
     end
+    def type=(t)
+      @type=t
+    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
+          if i==j
+            1.0
+          else
+            self[i,j].quo(Math::sqrt(self[i,i])*Math::sqrt(self[j,j]))
+          end
         }
       })
       matrix.extend CovariateMatrix 

data/lib/statsample/mle/logit.rb

@@ -1,12 +1,13 @@
 module Statsample
   module MLE
-    # Logit MLE estimation.
+    # Logit MLE estimation. 
+    # See Statsample::Regression for methods to generate a logit regression.
     # Usage:
     # 
     #   mle=Statsample::MLE::Logit.new
     #   mle.newton_raphson(x,y)
     #   beta=mle.parameters
-    #   likehood=mle.likehood(x,y,beta)
+    #   likehood=mle.likehood(x, y, beta)
     #   iterations=mle.iterations
     #         
     class Logit < BaseMLE

data/lib/statsample/mle/probit.rb

@@ -1,7 +1,9 @@
 module Statsample
   module MLE
     # Probit MLE estimation.
-    # Usage:
+    # See Statsample::Regression for methods to generate a probit regression.    
+    #
+    # == Usage:
     # 
     #   mle=Statsample::MLE::Probit.new
     #   mle.newton_raphson(x,y)
@@ -10,7 +12,7 @@ module Statsample
     #   iterations=mle.iterations
     class Probit < BaseMLE
       # F(B'Xi)
-      if  HAS_GSL
+      if  Statsample.has_gsl?
         # F(B'Xi)
         def f(b,x)
             p_bx=(x*b)[0,0] 
@@ -67,7 +69,7 @@ module Statsample
         raise "x.columns!=p.rows" if x.column_size!=b.row_size
         n = x.row_size
         k = x.column_size
-        if HAS_GSL
+        if Statsample.has_gsl?
           sum=GSL::Matrix.zeros(k)
         else
           sum=Matrix.zero(k)
@@ -76,12 +78,12 @@ module Statsample
           xi=Matrix.rows([x.row(i).to_a])
           fbx=f(b,xi)
           val=((ff(b,xi)**2) / (fbx*(1.0-fbx)))*xi.t*xi
-          if HAS_GSL
+          if Statsample.has_gsl?
             val=val.to_gsl
           end
           sum-=val
         end
-        if HAS_GSL
+        if Statsample.has_gsl?
           sum=sum.to_matrix
         end
         sum

data/lib/statsample/mle.rb

@@ -1,7 +1,9 @@
 module Statsample
-  # Module for MLE calculations.
+  # Module for generic MLE calculations.
   # Use subclass of BaseMLE for specific MLE model estimation.
-  # Usage:
+  # You should visit Statsample::Regression for method to perform fast
+  # regression analysis. 
+  # == Usage:
   # 
   #   mle=Statsample::MLE::Probit.new
   #   mle.newton_raphson(x,y)

data/lib/statsample/multiset.rb

@@ -24,53 +24,53 @@ module Statsample
         @datasets.size
     end
     def add_dataset(key,ds)
-        if(ds.fields!=@fields)
-        raise ArgumentError, "Dataset(#{ds.fields.to_s})must have the same fields of the Multiset(#{@fields})"
-        else
-            @datasets[key]=ds
-        end
+      if(ds.fields!=@fields)
+       raise ArgumentError, "Dataset(#{ds.fields.to_s})must have the same fields of the Multiset(#{@fields})"
+      else
+          @datasets[key]=ds
+      end
     end
     def sum_field(field)
-    @datasets.inject(0) {|a,da|
-    stratum_name=da[0]
-            vector=da[1][field]
-    val=yield stratum_name,vector
-    a+val
-    }
+      @datasets.inject(0) {|a,da|
+        stratum_name=da[0]
+        vector=da[1][field]
+        val=yield stratum_name,vector
+        a+val
+      }
     end
     def collect_vector(field)
-        @datasets.collect {|k,v|
-            yield k, v[field]
-        }
+      @datasets.collect {|k,v|
+        yield k, v[field]
+      }
     end
     def[](i)
-        @datasets[i]
-    end
+      @datasets[i]
     end
-    class StratifiedSample
-      class << self
-			# mean for an array of vectors
-			def mean(*vectors)
-				n_total=0
-				means=vectors.inject(0){|a,v|
-					n_total+=v.size
-					a+v.sum
-				}
-				means.to_f/n_total
-			end
-
+  end
+  class StratifiedSample
+    class << self
+      # mean for an array of vectors
+      def mean(*vectors)
+        n_total=0
+        means=vectors.inject(0){|a,v|
+          n_total+=v.size
+          a+v.sum
+        }
+        means.to_f/n_total
+      end
+      
       def standard_error_ksd_wr(es)
-          n_total=0
-          sum=es.inject(0){|a,h|
-              n_total+=h['N']
-              a+((h['N']**2 * h['s']**2) / h['n'].to_f)
-          }
-          (1.to_f / n_total)*Math::sqrt(sum)
+        n_total=0
+        sum=es.inject(0){|a,h|
+            n_total+=h['N']
+            a+((h['N']**2 * h['s']**2) / h['n'].to_f)
+        }
+        (1.to_f / n_total)*Math::sqrt(sum)
       end
       
       
       def variance_ksd_wr(es)
-          standard_error_ksd_wr(es)**2
+        standard_error_ksd_wr(es)**2
       end
       def calculate_n_total(es)
         es.inject(0) {|a,h| a+h['N'] }
@@ -85,7 +85,7 @@ module Statsample
       }
       end
       def standard_error_ksd_wor(es)
-          Math::sqrt(variance_ksd_wor(es))
+        Math::sqrt(variance_ksd_wor(es))
       end
       
       
@@ -101,26 +101,26 @@ module Statsample
       
       
       def standard_error_esd_wor(es)
-          Math::sqrt(variance_ksd_wor(es))
+        Math::sqrt(variance_ksd_wor(es))
       end
       # Based on http://stattrek.com/Lesson6/STRAnalysis.aspx
       def variance_esd_wr(es)
         n_total=calculate_n_total(es)
           sum=es.inject(0){|a,h|
-              val= ((h['s']**2 * h['N']**2) / h['n'].to_f)
-              a+val
+            val= ((h['s']**2 * h['N']**2) / h['n'].to_f)
+            a+val
           }
           (1.0/(n_total**2))*sum
       end
       def standard_error_esd_wr(es)
-          Math::sqrt(variance_esd_wr(es))
+        Math::sqrt(variance_esd_wr(es))
       end
       
       def proportion_variance_ksd_wor(es)
         n_total=calculate_n_total(es)
           es.inject(0){|a,h|
-              val= (((h['N'].to_f / n_total)**2 * h['p']*(1-h['p'])) / (h['n'])) * (1- (h['n'].to_f / h['N']))
-              a+val
+            val= (((h['N'].to_f / n_total)**2 * h['p']*(1-h['p'])) / (h['n'])) * (1- (h['n'].to_f / h['N']))
+            a+val
           }
       end
       def proportion_sd_ksd_wor(es)
@@ -141,19 +141,20 @@ module Statsample
       end
       
       def proportion_variance_esd_wor(es)
-      n_total=n_total=calculate_n_total(es)
-          
-          sum=es.inject(0){|a,h|
-              a=(h['N']**2 * (h['N']-h['n']) * h['p']*(1.0-h['p'])) / ((h['n']-1)*(h['N']-1))
-              a+val
-          }
-          Math::sqrt(sum) * (1.0/n_total**2)
+        n_total=n_total=calculate_n_total(es)
+        
+        sum=es.inject(0){|a,h|
+          a=(h['N']**2 * (h['N']-h['n']) * h['p']*(1.0-h['p'])) / ((h['n']-1)*(h['N']-1))
+          a+val
+        }
+        Math::sqrt(sum) * (1.0/n_total**2)
       end
       def proportion_sd_esd_wor(es)
           Math::sqrt(proportion_variance_ksd_wor(es))
       end
-		end
-  def initialize(ms,strata_sizes)
+    end
+    
+    def initialize(ms,strata_sizes)
       raise TypeError,"ms should be a Multiset" unless ms.is_a? Statsample::Multiset
       @ms=ms
       raise ArgumentError,"You should put a strata size for each dataset" if strata_sizes.keys.sort!=ms.datasets_names
@@ -161,104 +162,104 @@ module Statsample
       @population_size=@strata_sizes.inject(0) {|a,x| a+x[1]}
       @strata_number=@ms.n_datasets
       @sample_size=@ms.datasets.inject(0) {|a,x| a+x[1].cases}
-  end
-  # Number of strata
-  def strata_number
+    end
+    # Number of strata
+    def strata_number
       @strata_number
-  end
-  # Population size. Equal to sum of strata sizes
-  # Symbol: N<sub>h</sub>
-  def population_size
+    end
+    # Population size. Equal to sum of strata sizes
+    # Symbol: N<sub>h</sub>
+    def population_size
       @population_size
-  end
-  # Sample size. Equal to sum of sample of each stratum
-  def sample_size
+    end
+    # Sample size. Equal to sum of sample of each stratum
+    def sample_size
       @sample_size
-  end
-  # Size of stratum x
-  def stratum_size(h)
+    end
+    # Size of stratum x
+    def stratum_size(h)
       @strata_sizes[h]
-  end
-  def vectors_by_field(field)
+    end
+    def vectors_by_field(field)
       @ms.datasets.collect{|k,ds|
-          ds[field]
+        ds[field]
       }
-  end
-  # Population proportion based on strata
-  def proportion(field, v=1)
-  @ms.sum_field(field) {|s_name,vector|
-  stratum_ponderation(s_name)*vector.proportion(v)
-  }
-  end
-  # Stratum ponderation.
-  # Symbol: W\<sub>h\</sub>
-  def stratum_ponderation(h)
-    @strata_sizes[h].to_f / @population_size
-  end
-  alias_method :wh, :stratum_ponderation
-  
-  # Population mean based on strata
-  def mean(field)
-    @ms.sum_field(field) {|s_name,vector|
-    stratum_ponderation(s_name)*vector.mean
-    }
-  end
-  # Standard error with estimated population variance and without replacement.
-  # Source: Cochran (1972)
-  def standard_error_wor(field)
+    end
+    # Population proportion based on strata
+    def proportion(field, v=1)
+      @ms.sum_field(field) {|s_name,vector|
+      stratum_ponderation(s_name)*vector.proportion(v)
+      }
+    end
+    # Stratum ponderation.
+    # Symbol: W\<sub>h\</sub>
+    def stratum_ponderation(h)
+      @strata_sizes[h].to_f / @population_size
+    end
+    alias_method :wh, :stratum_ponderation
+    
+    # Population mean based on strata
+    def mean(field)
+      @ms.sum_field(field) {|s_name,vector|
+      stratum_ponderation(s_name)*vector.mean
+      }
+    end
+    # Standard error with estimated population variance and without replacement.
+    # Source: Cochran (1972)
+    def standard_error_wor(field)
       es=@ms.collect_vector(field) {|s_n, vector|
-          {'N'=>@strata_sizes[s_n],'n'=>vector.size, 's'=>vector.sds}
+        {'N'=>@strata_sizes[s_n],'n'=>vector.size, 's'=>vector.sds}
       }
       
       StratifiedSample.standard_error_esd_wor(es)
-  end
-  
-  # Standard error with estimated population variance and without replacement.
-  # Source: http://stattrek.com/Lesson6/STRAnalysis.aspx
-  
-  def standard_error_wor_2(field)
-  sum=@ms.sum_field(field) {|s_name,vector|
-          s_size=@strata_sizes[s_name]
-  (s_size**2 * (1-(vector.size.to_f / s_size)) * vector.variance_sample / vector.size.to_f)
-  }
+    end
+    
+    # Standard error with estimated population variance and without replacement.
+    # Source: http://stattrek.com/Lesson6/STRAnalysis.aspx
+    
+    def standard_error_wor_2(field)
+      sum=@ms.sum_field(field) {|s_name,vector|
+        s_size=@strata_sizes[s_name]
+      (s_size**2 * (1-(vector.size.to_f / s_size)) * vector.variance_sample / vector.size.to_f)
+      }
       (1/@population_size.to_f)*Math::sqrt(sum)
-  end
-  
-  def standard_error_wr(field)
+    end
+    
+    def standard_error_wr(field)
       es=@ms.collect_vector(field) {|s_n, vector|
-          {'N'=>@strata_sizes[s_n],'n'=>vector.size, 's'=>vector.sds}
+        {'N'=>@strata_sizes[s_n],'n'=>vector.size, 's'=>vector.sds}
       }
       
       StratifiedSample.standard_error_esd_wr(es)
-  end
-  def proportion_sd_esd_wor(field,v=1)
+    end
+    def proportion_sd_esd_wor(field,v=1)
       es=@ms.collect_vector(field) {|s_n, vector|
-          {'N'=>@strata_sizes[s_n],'n'=>vector.size, 'p'=>vector.proportion(v)}
+        {'N'=>@strata_sizes[s_n],'n'=>vector.size, 'p'=>vector.proportion(v)}
       }
       
       StratifiedSample.proportion_sd_esd_wor(es)
-  end
-  
-  def proportion_standard_error(field,v=1)
+    end
+    
+    def proportion_standard_error(field,v=1)
       prop=proportion(field,v)
       sum=@ms.sum_field(field) {|s_name,vector|
-          nh=vector.size
-          s_size=@strata_sizes[s_name]
-          (s_size**2 * (1-(nh/s_size)) * prop * (1-prop) / (nh -1 ))
+        nh=vector.size
+        s_size=@strata_sizes[s_name]
+        (s_size**2 * (1-(nh/s_size)) * prop * (1-prop) / (nh -1 ))
       }
       (1.quo(@population_size)) * Math::sqrt(sum)
-  end
-  # Cochran(1971), p. 150 
-  def variance_pst(field,v=1)
+    end
+    # Cochran(1971), p. 150 
+    def variance_pst(field,v=1)
       sum=@ms.datasets.inject(0) {|a,da|
-          stratum_name=da[0]
-          ds=da[1]
-          nh=ds.cases.to_f
-          s_size=@strata_sizes[stratum_name]
-          prop=ds[field].proportion(v)
-          a + (((s_size**2 * (s_size-nh)) / (s_size-1))*(prop*(1-prop) / (nh-1)))
+        stratum_name=da[0]
+        ds=da[1]
+        nh=ds.cases.to_f
+        s_size=@strata_sizes[stratum_name]
+        prop=ds[field].proportion(v)
+        a + (((s_size**2 * (s_size-nh)) / (s_size-1))*(prop*(1-prop) / (nh-1)))
       }
       (1/@population_size.to_f ** 2)*sum
-  end
+    end
   end
 end

data/lib/statsample/permutation.rb

@@ -14,7 +14,8 @@ module Statsample
   #  perm=Statsample::Permutation.new([0,0,1,1])
   #  => [[0,0,1,1],[0,1,0,1],[0,1,1,0],[1,0,0,1],[1,0,1,0],[1,1,0,0]]
   #
-  # Reference: http://www.cut-the-knot.org/do_you_know/AllPerm.shtml
+  # == Reference:
+  # * http://www.cut-the-knot.org/do_you_know/AllPerm.shtml
   class Permutation
     attr_reader :permutation_number
     def initialize(v)

data/lib/statsample/regression/binomial/logit.rb

@@ -1,11 +1,12 @@
 module Statsample
   module Regression
     module Binomial
-      # Logistic Regression
+      # Logistic Regression class.
+      # See Statsample::Regression::Binomial::BaseEngine for documentation
       class Logit < BaseEngine
         def initialize(ds,y_var)
-            model=Statsample::MLE::Logit.new
-            super(ds,y_var,model)
+          model=Statsample::MLE::Logit.new
+          super(ds,y_var,model)
         end
       end
     end

data/lib/statsample/regression/binomial/probit.rb

@@ -1,7 +1,8 @@
 module Statsample
   module Regression
     module Binomial
-      # Logistic Regression
+      # Probit Regression
+      # See Statsample::Regression::Binomial::BaseEngine for documentation
       class Probit < BaseEngine
         def initialize(ds,y_var)
           model=Statsample::MLE::Probit.new

data/lib/statsample/regression/binomial.rb

@@ -1,91 +1,72 @@
 module Statsample
   module Regression
     module Binomial
-      # Create a Logit model object.
-      # ds:: Dataset
-      # y::  Name of dependent vector
-      # Use
-      #   dataset=Statsample::CSV.read("data.csv")
-      #   y="y" 
-      #   lr=Statsample::Regression::Binomial.logit(dataset,y)
-      #   
-      def self.logit(ds,y_var)
-        Logit.new(ds,y_var)                
-      end
-      # Create a Probit model object.
-      # ds:: Dataset
-      # y::  Name of dependent vector
-      # Use
-      #   dataset=Statsample::CSV.read("data.csv")
-      #   y="y" 
-      #   lr=Statsample::Regression::Binomial.probit(dataset,y)
-      #   
-      
-      def self.probit(ds,y_var)
-        Probit.new(ds,y_var)                
-      end
       # Base Engine for binomial regression analysis.
-      # See Statsample::Regression::Binomial.logit() and
-      # Statsample::Regression::Binomial.probit for fast
-      # access methods.
-      # 
-      # Use:
-      #   dataset=Statsample::CSV.read("data.csv")
-      #   y="y" 
-      #   model=Statsample::MLE::Logit.new
-      #   lr=Statsample::Regression::Binomial::BaseEngine(dataset, y, model)
-    class BaseEngine
-      attr_reader :log_likehood, :iterations
-      def initialize(ds,y_var,model)
-      @ds=ds
-      @y_var=y_var
-      @dy=@ds[@y_var]
-      @ds_indep=ds.dup(ds.fields-[y_var])
-      constant=([1.0]*ds.cases).to_vector(:scale)
-      @ds_indep.add_vector("_constant",constant)
-      mat_x=@ds_indep.to_matrix
-      mat_y=@dy.to_matrix(:vertical)
-      @fields=@ds_indep.fields
-      @model=model
-      coeffs=model.newton_raphson(mat_x, mat_y)
-      @coeffs=assign_names(coeffs.column(0).to_a)
-      @iterations=model.iterations
-      @var_cov_matrix=model.var_cov_matrix
-      @log_likehood=model.log_likehood(mat_x, mat_y, coeffs)
-      end # init
-      # Coefficients standard error
-      def coeffs_se
-        out={}
-        @fields.each_index{|i|
+      # Use Statsample::Regression.logit and Statsample::Regression.probit
+      # for fast access methods.
+      #
+      # == Usage:
+      #  dataset=Statsample::CSV.read("data.csv")
+      #  y="y"
+      #  model=Statsample::MLE::Logit.new
+      #  lr=Statsample::Regression::Binomial::BaseEngine(dataset, y, model)
+      class BaseEngine
+        attr_reader :log_likehood, :iterations
+        # Parameters
+        # * ds: Dataset
+        # * y_var: Name of dependent variable
+        # * model: One of Statsample::Regression::Binomial classes
+        def initialize(ds,y_var,model)
+          @ds=ds
+          @y_var=y_var
+          @dy=@ds[@y_var]
+          @ds_indep=ds.dup(ds.fields-[y_var])
+          constant=([1.0]*ds.cases).to_vector(:scale)
+          @ds_indep.add_vector("_constant",constant)
+          mat_x=@ds_indep.to_matrix
+          mat_y=@dy.to_matrix(:vertical)
+          @fields=@ds_indep.fields
+          @model=model
+          coeffs=model.newton_raphson(mat_x, mat_y)
+          @coeffs=assign_names(coeffs.column(0).to_a)
+          @iterations=model.iterations
+          @var_cov_matrix=model.var_cov_matrix
+          @log_likehood=model.log_likehood(mat_x, mat_y, coeffs)
+        end # init
+        # Coefficients standard error
+        def coeffs_se
+          out={}
+          @fields.each_index{|i|
             f=@fields[i]
             out[f]=Math::sqrt(@var_cov_matrix[i,i])
-        }
-        out.delete("_constant")
-        out
-      end
-      # Constant value
-      def constant
-        @coeffs['_constant']
-      end
-      # Regression coefficients
-      def coeffs
-        c=@coeffs.dup
-        c.delete("_constant")
-        c
-      end
-      # Constant standard error
-      def constant_se
-        i=@fields.index :_constant
-        Math::sqrt(@var_cov_matrix[i,i])
-      end
-      def assign_names(c)
-        a={}
-        @fields.each_index do |i|
-          a[@fields[i]]=c[i]
+          }
+          out.delete("_constant")
+          out
+        end
+        # Value of constant on regression
+        def constant
+          @coeffs['_constant']
+        end
+        # Constant standard error
+        def constant_se
+          i=@fields.index :_constant
+          Math::sqrt(@var_cov_matrix[i,i])
+        end
+        # Regression coefficients
+        def coeffs
+          c=@coeffs.dup
+          c.delete("_constant")
+          c
+        end
+
+        def assign_names(c) # :nodoc:
+          a={}
+          @fields.each_index do |i|
+            a[@fields[i]]=c[i]
+          end
+          a
         end
-        a
-      end
       end # Base Engine
-    end # Dichotomic
+    end # Binomial
   end # Regression
 end # Stasample

data/lib/statsample/regression/multiple/baseengine.rb

@@ -110,7 +110,7 @@ module Statsample
           ds.each{|k,v|
           ds[k]=v.to_vector(:scale)
           }
-          lr=Multiple.listwise(ds.to_dataset,var)
+          lr=self.class.new(ds.to_dataset,var)
           1-lr.r2
         end
         # Tolerances for each coefficient