Refactor some code

lib/statsample-glm/glm.rb

@@ -27,10 +27,9 @@ module GLM
     def self.compute(data_set, dependent_column, method, opts={})
       opts[:method] = method
 
-      # TODO: Remove this const_get jugaad after 1.9.3 support is removed.
-
-      Kernel.const_get("Statsample").const_get("GLM").const_get("#{method.capitalize}").new data_set, 
-        dependent_column, opts
+      Kernel.const_get(
+        "Statsample::GLM::#{method.capitalize}"
+      ).new data_set, dependent_column, opts
     end
   end
 end

lib/statsample-glm/glm/base.rb

@@ -7,14 +7,15 @@
 module Statsample
   module GLM
     class Base
+      extend Gem::Deprecate
 
       def initialize ds, y, opts={}
         @opts   = opts
 
         set_default_opts_if_any
 
-        @data_set  = ds.dup(ds.vectors.to_a - [y])
         @dependent = ds[y]
+        @data_set = ds.delete_vector y
 
         if @opts[:constant]
           add_constant_vector @opts[:constant]
@@ -25,13 +26,12 @@ def initialize ds, y, opts={}
         algorithm = @opts[:algorithm].upcase
         method    = @opts[:method].capitalize
 
-        # TODO: Remove this const_get jugaad after 1.9.3 support is removed.
-
-        @regression = Kernel.const_get("Statsample").const_get("GLM")
-                            .const_get("#{algorithm}").const_get("#{method}")
-                            .new(@data_set, @dependent, @opts)
+        @regression = Object.const_get(
+                      "Statsample::GLM::#{algorithm}::#{method}"
+                      ).new @data_set, @dependent, @opts
       end
 
+
       # Returns the coefficients of trained model
       #
       # @param [Symbol] as_a Specifies the form of output
@@ -73,29 +73,33 @@ def coefficients as_a=:vector
       #   require 'statsample-glm'
       #   data_set = Daru::DataFrame.from_csv "spec/data/logistic.csv"
       #   glm  = Statsample::GLM.compute data_set, "y", :logistic, {constant: 1}
-      #   glm.standard_error
+      #   glm.standard_errors
       #     # #<Daru::Vector:25594060 @name = nil @metadata = {} @size = 3 >
       #     #                                     nil
       #     #                   0  0.4130813039878828
       #     #                   1  0.7194644911927432
       #     #                   2 0.40380565497038895
       #
-      def standard_error as_a=:vector  
+      def standard_errors as_a=:vector  
         case as_a
         when :hash
           se = {}
           @data_set.vectors.to_a.each_with_index do |f,i|
-            se[f.to_sym] = @regression.standard_error[i]
+            se[f.to_sym] = @regression.standard_errors[i]
           end
           se
         when :array
-          @regression.standard_error.to_a
+          @regression.standard_errors.to_a
         when :vector
-          @regression.standard_error
+          @regression.standard_errors
         else
           raise ArgumentError, "as_a has to be one of :array, :hash, or :vector"
         end
       end
+
+      # standard_error will be removed soon
+      alias :standard_error :standard_errors
+      deprecate :standard_error, :standard_errors, 2017, 1
 
       def iterations
         @regression.iterations
@@ -154,13 +158,17 @@ def residuals
       #   require 'statsample-glm'
       #   data_set = Daru::DataFrame.from_csv "spec/data/logistic.csv"
       #   glm  = Statsample::GLM.compute data_set, "y", :logistic, constant: 1
-      #   glm.degree_of_freedom
+      #   glm.degrees_of_freedom
       #     # => 47
       #
-      def degree_of_freedom
-        @regression.degree_of_freedom
+      def degrees_of_freedom
+        @regression.degrees_of_freedom
       end
 
+      # degrees_of_freedom will be removed soon
+      alias :degree_of_freedom :degrees_of_freedom
+      deprecate :degree_of_freedom, :degrees_of_freedom, 2017, 1
+
       # Returns the optimal value of the log-likelihood function when using MLE algorithm.
       # The optimal value is the value of the log-likelihood function at the MLE solution.
       #

lib/statsample-glm/glm/irls/base.rb

@@ -3,8 +3,8 @@ module GLM
     module IRLS
       class Base
 
-        attr_reader :coefficients, :standard_error, :iterations, 
-          :fitted_mean_values, :residuals, :degree_of_freedom
+        attr_reader :coefficients, :standard_errors, :iterations, 
+          :fitted_mean_values, :residuals, :degrees_of_freedom
 
         def initialize data_set, dependent, opts={}
           @data_set  = data_set.to_matrix
@@ -61,13 +61,13 @@ def irls
 
           @coefficients       = create_vector(b.column_vectors[0])
           @iterations         = max_iter
-          @standard_error     = create_vector(hessian(@data_set,b).inverse
+          @standard_errors     = create_vector(hessian(@data_set,b).inverse
                                                                   .diagonal
                                                                   .map{ |x| -x}
                                                                   .map{ |y| Math.sqrt(y) })
           @fitted_mean_values = create_vector measurement(@data_set,b).to_a.flatten
           @residuals          = @dependent - @fitted_mean_values
-          @degree_of_freedom  = @dependent.count - @data_set.column_size
+          @degrees_of_freedom  = @dependent.count - @data_set.column_size
         end
 
         def create_vector arr

lib/statsample-glm/glm/mle/base.rb

@@ -4,7 +4,7 @@ module GLM
     module MLE
       class Base
           attr_reader :coefficients, :iterations, 
-            :fitted_mean_values, :residuals, :degree_of_freedom,
+            :fitted_mean_values, :residuals, :degrees_of_freedom,
             :log_likelihood
 
         MIN_DIFF_PARAMETERS=1e-2
@@ -27,15 +27,15 @@ def initialize data_set, dependent, opts
           @log_likelihood = _log_likelihood x, y, @coefficients
           @fitted_mean_values = create_vector measurement(x, @coefficients).to_a.flatten
           @residuals = @dependent - @fitted_mean_values
-          @degree_of_freedom  = @dependent.count - x.column_size
+          @degrees_of_freedom  = @dependent.count - x.column_size
 
           # This jugad is done because the last vector index for Normal is sigma^2
           # which we dont want to return to the user.
           @coefficients =  create_vector(self.is_a?(Statsample::GLM::MLE::Normal) ? 
             @coefficients.to_a.flatten[0..-2] : @coefficients.to_a.flatten)
         end
 
-        def standard_error
+        def standard_errors
           out = []
 
           @data_set.vectors.to_a.each_index do |i|

spec/logistic_spec.rb

@@ -54,15 +54,15 @@
     end
 
     it "reports correct standard deviations as a Daru::Vector" do
-      expect_similar_vector(@glm.standard_error, [0.4390, 0.4270, 0.3819,1.9045], 0.001)
+      expect_similar_vector(@glm.standard_errors, [0.4390, 0.4270, 0.3819,1.9045], 0.001)
     end
 
     it "reports correct regression coefficients as an array" do
       expect_similar_array(@glm.coefficients(:array), [0.3270, 0.8147, -0.4031,-5.3658], 0.001)
     end
 
     it "reports correct standard deviations as an array" do
-      expect_similar_array(@glm.standard_error(:array), [0.4390, 0.4270, 0.3819,1.9045], 0.001)
+      expect_similar_array(@glm.standard_errors(:array), [0.4390, 0.4270, 0.3819,1.9045], 0.001)
     end
 
     it "reports correct regression coefficients as a hash" do
@@ -71,7 +71,7 @@
     end
 
     it "reports correct standard deviations as a hash" do
-      expect_similar_hash(@glm.standard_error(:hash), {:a => 0.4390, :b => 0.4270, :c => 0.3819,
+      expect_similar_hash(@glm.standard_errors(:hash), {:a => 0.4390, :b => 0.4270, :c => 0.3819,
                                                        :constant => 1.9045}, 0.001)
     end