Merge pull request #43 from michaelphines/chi-square-vector

Add missing test and implementation for doing a chi-squared test with vectors

Author: agarie

lib/statsample/test/chisquare.rb

@@ -1,43 +1,73 @@
 module Statsample
   module Test
     module ChiSquare
-      class WithMatrix
+      module Shared
         attr_reader :df
         attr_reader :value
+
+        def to_f
+          @value
+        end
+
+        def chi_square
+          @value
+        end
+
+        def probability
+          1-Distribution::ChiSquare.cdf(@value.to_f,@df)
+        end
+      end
+
+      class WithMatrix
+        include Statsample::Test::ChiSquare::Shared
+
         def initialize(observed, expected=nil)
           @observed=observed
           @expected=expected or calculate_expected
           raise "Observed size!=expected size" if @observed.row_size!=@expected.row_size or @observed.column_size!=@expected.column_size
           @df=(@observed.row_size-1)*(@observed.column_size-1)
           @value=compute_chi
         end
+
         def calculate_expected
           sum=@observed.total_sum
           @expected=Matrix.rows( @observed.row_size.times.map {|i|
             @observed.column_size.times.map {|j|
               (@observed.row_sum[i].quo(sum) * @observed.column_sum[j].quo(sum))*sum
             }
-          })          
-        end
-        def to_f
-          @value
-        end
-        def chi_square
-          @value
-        end
-        def probability
-          1-Distribution::ChiSquare.cdf(@value.to_f,@df)
+          })
         end
+
         def compute_chi
-            sum=0
-            (0...@observed.row_size).each {|i|
-              (0...@observed.column_size).each {|j|
+          sum=0
+          (0...@observed.row_size).each {|i|
+            (0...@observed.column_size).each {|j|
               sum+=((@observed[i, j] - @expected[i,j])**2).quo(@expected[i,j])
-              }
             }
-            sum
+          }
+          sum
+        end
+      end
+
+      class WithVector
+        include Statsample::Test::ChiSquare::Shared
+
+        def initialize(observed, expected)
+          @observed = observed
+          @expected = expected
+          raise "Observed size!=expected size" if @observed.size!=@expected.size
+          @df = @observed.size - 1
+          @value = compute_chi
+        end
+
+        def compute_chi
+          sum=0
+          (0...@observed.size).each {|i|
+            sum+=((@observed[i] - @expected[i])**2).quo(@expected[i])
+          }
+          sum
         end
       end
     end
   end
-end
+end

test/test_stest.rb

@@ -23,6 +23,20 @@ def test_chi_square_matrix_only_observed
     assert_equal(6, chi.df)
   end
 
+  def test_chi_square_vector
+    observed = Vector[20,30,15]
+    expected = Vector[20,20,20]
+    assert_nothing_raised do
+      Statsample::Test.chi_square(observed, expected)
+    end
+    chi = Statsample::Test.chi_square(observed, expected)
+
+    assert_in_delta(6.25, chi.chi_square, 0.0001)
+    assert_in_delta(0.04393, chi.probability, 0.00001)
+
+    assert_equal(2, chi.df)
+  end
+
   def test_u_mannwhitney
     a = Daru::Vector.new([1, 2, 3, 4, 5, 6])
     b = Daru::Vector.new([0, 5, 7, 9, 10, 11])