Justin extensions

.gitignore

@@ -1,4 +1,3 @@
-doc.yaml
 *.swp
 *.rbc
 coverage
@@ -10,4 +9,21 @@ doc
 examples/images/*
 examples/*.html
 web/upload_task.rb
-.idea
+
+# From standard gemfile
+*.gem
+.bundle
+.config
+.yardoc
+Gemfile.lock
+InstalledFiles
+_yardoc
+coverage
+doc/
+lib/bundler/man
+pkg
+rdoc
+spec/reports
+test/tmp
+test/version_tmp
+tmp

Gemfile

@@ -1,18 +1,3 @@
 source "https://www.rubygems.org"
-gem 'minitest'
-gem 'rdoc'
-gem 'mocha', '0.14.0' #:require=>'mocha/setup'
-gem 'shoulda','3.5.0'
-gem 'shoulda-matchers','2.2.0'
-gem 'hoe'
-#gem 'bio-statsample-timeseries'
-gem 'reportbuilder'
-gem 'dirty-memoize'
-gem 'distribution'
-gem 'extendmatrix'
-gem 'minimization'
-gem 'rserve-client'
-gem 'rubyvis'
-gem 'spreadsheet'
-gem 'rb-gsl'
-gem 'awesome_print'
+
+gemspec

Rakefile

@@ -7,6 +7,7 @@ require 'rubygems'
 require 'statsample'
 require 'hoe'
 require 'rdoc'
+require "bundler/gem_tasks"
 
 Hoe.plugin :git
 Hoe.plugin :doofus

examples/correlation_matrix.rb

@@ -11,6 +11,12 @@
     'd'=>rnorm(samples))
   cm=cor(ds)
   summary(cm)
+
+  cp = corp(ds)
+  summary(cp)
+
+  cov = cov(ds)
+  summary(cov)
 end
 
 if __FILE__==$0

lib/statsample/bivariate.rb

@@ -71,6 +71,12 @@ def t_pearson(v1,v2)
       # giving r and vector size
       # Source : http://faculty.chass.ncsu.edu/garson/PA765/correl.htm
       def t_r(r,size)
+        raise "In computing value for t test for a pearson correlation, invalid size of series: #{size}" if size <= 2
+        raise "In computing value for t test for a pearson correlation, invalid value of r: #{r}" if r >= 1.1
+        if r >= 1
+          puts "StatSample::Bivariate#t_r: got an R value > 1 (#{r}), so substituting 0.999999"
+          r = 0.999999
+        end
         r * Math::sqrt(((size)-2).to_f / (1 - r**2))
       end
       # Retrieves the probability value (a la SPSS)
@@ -251,7 +257,11 @@ def correlation_probability_matrix(ds, tails=:both)
             (row==col or ds[row].type!=:scale or ds[col].type!=:scale) ? nil : prop_pearson(t_pearson(ds[row],ds[col]), v1a.size, tails)
           end
         end
-        Matrix.rows(rows)
+        m = Matrix.rows(rows)
+        m.extend(Statsample::CovariateMatrix)
+        m.fields=ds.fields
+        m.name = "Correlation Probability"
+        m
       end
 
       # Spearman ranked correlation coefficient (rho) between 2 vectors

lib/statsample/matrix.rb

@@ -27,14 +27,14 @@ def to_dataset
   if defined? :eigenpairs
     alias_method :eigenpairs_ruby, :eigenpairs
   end
-  
+
   if Statsample.has_gsl?
     # Optimize eigenpairs of extendmatrix module using gsl
     def eigenpairs
       to_gsl.eigenpairs
     end
   end
-  
+
   def eigenvalues
     eigenpairs.collect {|v| v[0]}
   end
@@ -44,11 +44,11 @@ def eigenvectors
   def eigenvectors_matrix
     Matrix.columns(eigenvectors)
   end
-
-
 
-
-
+
+
+
+
   def to_gsl
     out=[]
     self.row_size.times{|i|
@@ -76,7 +76,7 @@ class Matrix
     def to_gsl
       self
     end
-    
+
     def to_dataset
       f = (self.respond_to? :fields_y) ? fields_y : column_size.times.map {|i| _("VAR_%d") % (i+1) }
       ds=Statsample::Dataset.new(f)
@@ -91,7 +91,7 @@ def to_dataset
       ds.name=self.name if self.respond_to? :name
       ds
     end
-    
+
     def row_size
       size1
     end
@@ -110,18 +110,18 @@ def eigenvalues
     def eigenvectors
       eigenpairs.collect {|v| v[1]}
     end
-    
+
     # Matrix sum of squares
     def mssq
       sum=0
       to_v.each {|i| sum+=i**2}
       sum
     end
-    
+
     def eigenvectors_matrix
       eigval, eigvec= GSL::Eigen.symmv(self)
       GSL::Eigen::symmv_sort(eigval, eigvec, GSL::Eigen::SORT_VAL_DESC)
-      eigvec 
+      eigvec
     end
     def eigenpairs
       eigval, eigvec= GSL::Eigen.symmv(self)
@@ -130,7 +130,7 @@ def eigenpairs
         [eigval[i],eigvec.get_col(i)]
       }
     end
-    
+
     #def eigenpairs_ruby
     #  self.to_matrix.eigenpairs_ruby
     #end
@@ -158,7 +158,7 @@ def total_sum
 module Statsample
   # Module to add names to X and Y fields
   module NamedMatrix
-    include Summarizable  
+    include Summarizable
 
     def fields
     raise "Should be square" if !square?
@@ -178,10 +178,10 @@ def fields_y=(v)
     @fields_y=v
     end
     def fields_x
-    @fields_x||=row_size.times.collect {|i| _("X%d") % i} 
+    @fields_x||=row_size.times.collect {|i| _("X%d") % i}
     end
     def fields_y
-    @fields_y||=column_size.times.collect {|i| _("Y%d") % i} 
+    @fields_y||=column_size.times.collect {|i| _("Y%d") % i}
     end
 
     def name
@@ -195,7 +195,7 @@ def get_new_name
       @@named_matrix+=1
       _("Matrix %d") % @@named_matrix
     end
-    
+
   end
   # Module to add method for variance/covariance and correlation matrices
   # == Usage
@@ -209,15 +209,19 @@ module CovariateMatrix
     # Get type of covariate matrix. Could be :covariance or :correlation
     def _type
       if row_size==column_size
-        if row_size.times.find {|i| self[i,i]!=1.0}
-          :covariance
-        else
-          :correlation
+        @type ||= begin
+          if row_size.times.find { |i| self[i, i] == 1.0 }
+            :correlation
+          elsif row_size.times.find { |i| self[i, i].nil? }
+            :correlation_probability
+          else
+            :covariance
+          end
         end
       else
         @type
       end
-      
+
     end
     def _type=(t)
       @type=t
@@ -233,7 +237,7 @@ def correlation
             end
           }
         })
-        matrix.extend CovariateMatrix 
+        matrix.extend CovariateMatrix
         matrix.fields_x=fields_x
         matrix.fields_y=fields_y
         matrix._type=:correlation
@@ -242,19 +246,19 @@ def correlation
         self
       end
     end
-    
-    
+
+
     # Get variance for field k
     # 
     def variance(k)
       submatrix([k])[0,0]
     end
-    
+
     def get_new_name
       @@covariatematrix+=1
       _("Covariate matrix %d") % @@covariatematrix
     end
-    
+
     # Select a submatrix of factors. If you have a correlation matrix
     # with a, b and c, you could obtain a submatrix of correlations of
     # a and b, b and c or a and b
@@ -276,24 +280,24 @@ def submatrix(rows,columns=nil)
       raise ArgumentError, "rows shouldn't be empty" if rows.respond_to? :size and rows.size==0
       columns||=rows
       # Convert all fields on index
-      row_index=rows.collect {|v| 
+      row_index=rows.collect {|v|
         r=v.is_a?(Numeric) ? v : fields_x.index(v)
         raise "Index #{v} doesn't exists on matrix" if r.nil?
         r
       }
-      column_index=columns.collect {|v| 
+      column_index=columns.collect {|v|
         r=v.is_a?(Numeric) ? v : fields_y.index(v)
         raise "Index #{v} doesn't exists on matrix" if r.nil?
         r
       }
-      
-      
+
+
       fx=row_index.collect {|v| fields_x[v]}
       fy=column_index.collect {|v| fields_y[v]}
-        
+
       matrix= Matrix.rows(row_index.collect {|i|
         row=column_index.collect {|j| self[i,j]}})
-      matrix.extend CovariateMatrix 
+      matrix.extend CovariateMatrix
       matrix.fields_x=fx
       matrix.fields_y=fy
       matrix._type=_type

lib/statsample/shorthand.rb

@@ -40,9 +40,15 @@ def names(ds)
     def cor(ds)
       Statsample::Bivariate.correlation_matrix(ds)
     end
+
+    # Create a correlation probability matrix from a dataset
+    def corp(ds)
+      Statsample::Bivariate.correlation_probability_matrix(ds)
+    end
+
     # Create a variance/covariance matrix from a dataset
     def cov(ds)
-      Statsample::Bivariate.covariate_matrix(ds)
+      Statsample::Bivariate.covariance_matrix(ds)
     end
     # Create a Statsample::Vector
     # Analog to R's c