scoruby 0.2.7 → 0.2.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: aba708affa4e656a44a43d1f24b71b696760b7bd
-  data.tar.gz: d52253a2d308e3d7f7749f379feae9757e866887
+  metadata.gz: f78c07d033cba7c4f13531ae30e51f9c00b33d11
+  data.tar.gz: d487653a5f8604e669db89a8766b4e84d3e1950b
 SHA512:
-  metadata.gz: 7f1dcf1c79cea6278c5a584c7390acaf275104fde029e83514d3cbf39859c840944d3893076c658c6eca9506add2c192525f5fbf6f6c65e1e8871ff5ddd6429f
-  data.tar.gz: e9d4f29650d95e74ae6e45c5fd5a4ef8435d552b792f558c9e15c07ce6d4d02f2bbe64ae34d07cfcbc55a4afe0f554df2adb79b25d0a1dfc69d611cc3b660a8a
+  metadata.gz: d64f07f42472a9ae54435d26d738f05381a3ae2542c4ad2fdfdf879696e0af7c6bd8df6d946fb4aafed2e316fcc00c09c75b164c6a1d56b11ea5cd2ed48b82a5
+  data.tar.gz: 5ffde48fb242ef57f3c06f089aecfe73b5b9b8d9038a75caae516a368619aba3f1d597ddf664783346c538f26fbe9756efc02d20036900d2d94abd0fb7219204

data/.rubocop.yml

@@ -0,0 +1,8 @@
+Metrics/BlockLength:
+  ExcludedMethods: ['describe', 'context']
+AllCops:
+  Excludes:
+    - '**'
+  TargetRubyVersion: 2.4
+Documentation:
+  Enabled: false

data/.travis.yml

@@ -1,4 +1,6 @@
 language: ruby
 rvm:
-  - 2.4.1
-before_install: gem install bundler -v 1.10.5
+  - 2.2.8
+  - 2.3.5
+  - 2.4.2
+before_install: gem install bundler -v 1.15.4

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    scoruby (0.2.6)
+    scoruby (0.2.7)
       nokogiri (~> 1.7)
 
 GEM
@@ -18,9 +18,9 @@ GEM
     docile (1.1.5)
     json (2.1.0)
     method_source (0.8.2)
-    mini_portile2 (2.2.0)
-    nokogiri (1.8.0)
-      mini_portile2 (~> 2.2.0)
+    mini_portile2 (2.3.0)
+    nokogiri (1.8.1)
+      mini_portile2 (~> 2.3.0)
     pry (0.10.3)
       coderay (~> 1.1.0)
       method_source (~> 0.8.1)

data/README.md

@@ -7,12 +7,15 @@
 
 Ruby scoring API for Predictive Model Markup Language (PMML).
 
-Currently supports Decision Tree, Random Forest Naive Bayes and Gradient Boosted Models.
+Currently supports Decision Tree, Random Forest, Naive Bayes and Gradient Boosted Models.
 
 Will be happy to implement new models by demand, or assist with any other issue.
 
 Contact me here or at aschers@gmail.com.
 
+[Tutorial - Deploy Machine Learning Models from R Research to Ruby Production with PMML](https://medium.com/@aschers/deploy-machine-learning-models-from-r-research-to-ruby-go-production-with-pmml-b41e79445d3d)
+
+
 ## Installation
 
 Add this line to your application's Gemfile:
@@ -37,7 +40,7 @@ Or install it yourself as:
 
 ```ruby
 
-random_forest = Scoruby.get_model 'titanic_rf.pmml'
+random_forest = Scoruby.load_model 'titanic_rf.pmml'
 features =  {
         Sex: 'male',
         Parch: 0,
@@ -66,7 +69,7 @@ random_forest.decisions_count(features)
 
 ```ruby
 
-gbm = Scoruby.get_model 'gbm.pmml'
+gbm = Scoruby.load_model 'gbm.pmml'
 
 features =  {
         Sex: 'male',
@@ -87,17 +90,17 @@ gbm.score(features)
 ### Decision Tree
 
 ```ruby
-decision_tree = Scoruby.get_model 'decision_tree.pmml'
+decision_tree = Scoruby.load_model 'decision_tree.pmml'
 features = { f1 : v1, ... } 
 decision_tree.decide(features)
 
 => #<Decision:0x007fc232384180 @score="0", @score_distribution={"0"=>"0.999615579933873", "1"=>"0.000384420066126561"}>
 ```
 
-### Naibve Bayes
+### Naive Bayes
 
 ```ruby
-naive_bayes = Scoruby.get_model 'naive_bayes.pmml'
+naive_bayes = Scoruby.load_model 'naive_bayes.pmml'
 features = { f1: v1 , ... } 
 naive_bayes.lvalues(features)
 naive_bayes.score(features, 'l1')

data/bin/console

@@ -1,7 +1,8 @@
 #!/usr/bin/env ruby
+# frozen_string_literal: true
 
-require "bundler/setup"
-require "scoruby"
+require 'bundler/setup'
+require 'scoruby'
 
 # You can add fixtures and/or initialization code here to make experimenting
 # with your gem easier. You can also use a different console, if you like.
@@ -10,5 +11,5 @@ require "scoruby"
 # require "pry"
 # Pry.start
 
-require "pry"
+require 'pry'
 Pry.start

data/lib/scoruby/decision.rb

@@ -1,6 +1,7 @@
+# frozen_string_literal: true
+
 module Scoruby
   class Decision
-
     attr_reader :score, :score_distribution
 
     def initialize(score, score_distributions)
@@ -8,10 +9,11 @@ module Scoruby
       return if score_distributions.empty?
 
       @score_distribution = {}
-      score_distributions.each {|score_distribution|
-        attributes                                    = score_distribution.attributes
-        @score_distribution[attributes['value'].to_s] = attributes['probability'].to_s
-      }
+      score_distributions.each do |score_distribution|
+        value = score_distribution.attributes['value'].to_s
+        probability = score_distribution.attributes['probability'].to_s
+        @score_distribution[value] = probability
+      end
     end
   end
-end
+end

data/lib/scoruby/features.rb

@@ -1,6 +1,7 @@
+# frozen_string_literal: true
+
 module Scoruby
   class Features
-
     attr_reader :formatted
 
     def initialize(features)
@@ -8,11 +9,11 @@ module Scoruby
     end
 
     def format_booleans(features)
-      features.map {|k, v|
+      features.map do |k, v|
         features[k] = 'f' if v == false
         features[k] = 't' if v == true
-      }
+      end
       features
     end
   end
-end
+end

data/lib/scoruby/models/decision_tree.rb

@@ -1,9 +1,10 @@
+# frozen_string_literal: true
+
 require 'scoruby/node'
 
 module Scoruby
   module Models
     class DecisionTree
-
       attr_reader :root
 
       def initialize(tree_xml)
@@ -31,8 +32,9 @@ module Scoruby
       end
 
       def didnt_step?(curr, prev)
-        return false if (prev.pred != curr.pred)
-        Scoruby.logger.error "Null tree: #{@id}, bad feature: #{curr.children[0].pred.field }"
+        return false if prev.pred != curr.pred
+        feature = curr.children[0].pred.field
+        Scoruby.logger.error "Null tree: #{@id}, bad feature: #{feature}"
         true
       end
     end

data/lib/scoruby/models/gbm.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'scoruby/models/decision_tree'
 require 'scoruby/features'
 
@@ -8,10 +10,10 @@ module Scoruby
       CONST_XPATH      = '//Target/@rescaleConstant'
 
       def initialize(xml)
-        @decision_trees = xml.xpath(GBM_FOREST_XPATH).collect {|xml_tree|
+        @decision_trees = xml.xpath(GBM_FOREST_XPATH).map do |xml_tree|
           DecisionTree.new(xml_tree)
-        }
-        @const          = Float(xml.xpath(CONST_XPATH).to_s)
+        end
+        @const = Float(xml.xpath(CONST_XPATH).to_s)
       end
 
       def tree_count
@@ -20,11 +22,11 @@ module Scoruby
 
       def score(features)
         formatted_features = Features.new(features).formatted
-        x                  = @decision_trees.map {|dt|
-          score = dt.decide(formatted_features).score
-          score.to_s.to_f
-        }.reduce(:+) + @const
-        Math.exp(x) / (1 + Math.exp(x))
+        scores = @decision_trees.map do |dt|
+          dt.decide(formatted_features).score.to_s.to_f
+        end
+        sum = scores.reduce(:+) + @const
+        Math.exp(sum) / (1 + Math.exp(sum))
       end
     end
   end

data/lib/scoruby/models/naive_bayes/model.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'scoruby/models/naive_bayes/model_data'
 require 'forwardable'
 
@@ -6,7 +8,8 @@ module Scoruby
     module NaiveBayes
       class Model
         extend Forwardable
-        def_delegators :@model_data, :threshold, :labels, :numerical_features, :category_features
+        def_delegators :@model_data, :threshold, :labels, :numerical_features,
+                       :category_features
 
         def initialize(xml)
           @model_data = ModelData.new(xml)
@@ -35,28 +38,35 @@ module Scoruby
         end
 
         def calc_label_feature_values(features)
-          labels.each do |label, _|
+          labels.each_key do |label|
             features.each do |feature_name, feature_value|
-              label_value = calc_category(feature_name, feature_value, label)
-              label_value ||= calc_numerical(feature_name, feature_value, label)
+              label_value = category(feature_name, feature_value, label)
+              label_value ||= numerical(feature_name, feature_value, label)
               labels[label][feature_name] = label_value if label_value
             end
           end
         end
 
-        def calc_category(feature_name, feature_value, label)
-          return unless category_features[feature_name] && category_features[feature_name][feature_value]
-          value_count = category_features[feature_name][feature_value][label].to_f
-          overall_count = category_features[feature_name].sum { |_, value| value[label].to_f }
+        def category(feature_name, feature_value, label)
+          model_feature = category_features[feature_name]
+          return unless model_feature && model_feature[feature_value]
+          value_count = model_feature[feature_value][label].to_f
+          overall_count = model_feature.map { |_, value| value[label].to_f }
+                                       .reduce(0, :+)
           value_count / overall_count
         end
 
-        def calc_numerical(feature_name, feature_value, label)
-          return unless numerical_features[feature_name] && numerical_features[feature_name][label]
-          variance = numerical_features[feature_name][label][:variance].to_f
-          mean = numerical_features[feature_name][label][:mean].to_f
-          feature_value = feature_value.to_f
-          Math.exp(-(feature_value - mean)**2 / (2 * variance)) / Math.sqrt(2 * Math::PI * variance)
+        def numerical(feature_name, feature_value, label)
+          model_feature = numerical_features[feature_name]
+          return unless model_feature && model_feature[label]
+          calc_numerical(feature_value.to_f,
+                         model_feature[label][:mean].to_f,
+                         model_feature[label][:variance].to_f)
+        end
+
+        def calc_numerical(feature_value, mean, variance)
+          Math.exp(-(feature_value - mean)**2 / (2 * variance)) /
+            Math.sqrt(2 * Math::PI * variance)
         end
       end
     end

data/lib/scoruby/models/naive_bayes/model_data.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Scoruby
   module Models
     module NaiveBayes
@@ -14,15 +16,17 @@ module Scoruby
         private
 
         def fetch_threshold
-          @threshold = @xml.xpath('//NaiveBayesModel').attr('threshold').value.to_f
+          @threshold = @xml.xpath('//NaiveBayesModel').attr('threshold')
+                           .value.to_f
         end
 
         def fetch_features_data
           @category_features = {}
           @numerical_features = {}
           @xml.xpath('//BayesInput').each do |feature|
-            @category_features[feature.attr('fieldName').to_sym] = fetch_category_feature(feature)
-            @numerical_features[feature.attr('fieldName').to_sym] = fetch_numerical_feature(feature)
+            field_name = feature.attr('fieldName').to_sym
+            @category_features[field_name] = fetch_category_feature(feature)
+            @numerical_features[field_name] = fetch_numerical_feature(feature)
           end
         end
 
@@ -33,14 +37,14 @@ module Scoruby
             @labels[l.attr('value')] = { 'count': l.attr('count').to_f }
           end
         end
-        
+
         def fetch_numerical_feature(feature)
           return unless feature.child.name == 'TargetValueStats'
           features_data = {}
           feature.child.children.each do |child|
             features_data[child.attr('value').strip] = {
-                mean: child.child.attr('mean'),
-                variance: child.child.attr('variance')
+              mean: child.child.attr('mean'),
+              variance: child.child.attr('variance')
             }
           end
           features_data
@@ -65,4 +69,4 @@ module Scoruby
       end
     end
   end
-end
+end

data/lib/scoruby/models/random_forest.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Scoruby
   module Models
     class RandomForest
@@ -5,26 +7,39 @@ module Scoruby
 
       def initialize(xml)
         xml_trees       = xml.xpath(RF_FOREST_XPATH)
-        @decision_trees = xml_trees.collect {|xml_tree|
+        @decision_trees = xml_trees.map do |xml_tree|
           DecisionTree.new(xml_tree)
+        end
+      end
+
+      def predict(features)
+        decisions_count = decisions_count(features)
+        decision = decisions_count.max_by { |_, v| v }
+        {
+          label: decision[0],
+          score: decision[1] / decisions_count.values.reduce(0, :+).to_f
         }
       end
 
       def decisions_count(features)
         formatted_features = Features.new(features).formatted
-        decisions          = @decision_trees.collect {|decision_tree|
+        decisions = traverse_trees(formatted_features)
+        aggregate_decisions(decisions)
+      end
+
+      private
+
+      def traverse_trees(formatted_features)
+        @decision_trees.map do |decision_tree|
           decision_tree.decide(formatted_features).score
-        }
-        decisions.inject(Hash.new(0)) {|h, e| h[e] += 1; h}
+        end
       end
 
-      def predict(features)
-        decisions_count = decisions_count(features)
-        {
-            label: decisions_count.max_by {|_, v| v}[0],
-            score: decisions_count.max_by {|_, v| v}[1] / decisions_count.values.reduce(0, :+).to_f
-        }
+      def aggregate_decisions(decisions)
+        decisions.each_with_object(Hash.new(0)) do |score, counts|
+          counts[score] += 1
+        end
       end
     end
   end
-end
+end

data/lib/scoruby/models_factory.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'scoruby/models/decision_tree'
 require 'scoruby/models/gbm'
 require 'scoruby/models/random_forest'

data/lib/scoruby/node.rb

@@ -1,19 +1,18 @@
+# frozen_string_literal: true
+
 require 'scoruby/predicate_factory'
 require 'scoruby/decision'
 
 module Scoruby
   class Node
-
     attr_reader :decision, :pred, :children
 
     def initialize(xml)
       children = xml.children
-
+      distributions = children.select { |c| c.name == 'ScoreDistribution' }
       @decision = Decision.new(xml.attribute('score').to_s,
-        children.select {|c| c.name == 'ScoreDistribution'})
-
+                               distributions)
       children = remove_nodes(children)
-      
       @pred     = PredicateFactory.for(children[0])
       @children = children_nodes(children)
     end
@@ -26,11 +25,11 @@ module Scoruby
 
     def children_nodes(children)
       children.select { |c| c.name == 'Node' }
-        .map { |child| Node.new(child) }
+              .map { |child| Node.new(child) }
     end
 
     def remove_nodes(children)
-      children.reject {|c| %w(Extension ScoreDistribution).include? c.name}
+      children.reject { |c| %w[Extension ScoreDistribution].include? c.name }
     end
   end
-end
+end

data/lib/scoruby/predicate_factory.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'scoruby/predicates/compound_predicate'
 require 'scoruby/predicates/simple_predicate'
 require 'scoruby/predicates/simple_set_predicate'
@@ -6,14 +8,11 @@ require 'scoruby/predicates/false_predicate'
 
 module Scoruby
   class PredicateFactory
-
     def self.for(pred_xml)
-      return Predicates::SimplePredicate.new(pred_xml) if pred_xml.name == 'SimplePredicate'
-      return Predicates::SimpleSetPredicate.new(pred_xml) if pred_xml.name == 'SimpleSetPredicate'
-      return Predicates::CompoundPredicate.new(pred_xml) if pred_xml.name == 'CompoundPredicate'
       return Predicates::TruePredicate.new if pred_xml.name == 'True'
       return Predicates::FalsePredicate.new if pred_xml.name == 'False'
+      predicate = Object.const_get("Scoruby::Predicates::#{pred_xml.name}")
+      predicate.new pred_xml
     end
   end
 end
-

data/lib/scoruby/predicates/compound_predicate.rb

@@ -1,7 +1,8 @@
+# frozen_string_literal: true
+
 module Scoruby
   module Predicates
     class CompoundPredicate
-
       attr_reader :field
 
       def initialize(pred_xml)
@@ -21,24 +22,28 @@ module Scoruby
         and?(features) if @boolean_operator == 'and'
       end
 
-      def is_missing?(features)
-        @field.any? {|f| !features.keys.include?(f)}
+      def missing?(features)
+        @field.any? { |f| !features.keys.include?(f) }
       end
 
       private
 
       def surrogate?(features)
-        return @predicates[1].true?(features) if @predicates[0].is_missing?(features)
+        return @predicates[1].true?(features) if first_missing?(features)
         @predicates[0].true?(features)
       end
 
+      def first_missing?(features)
+        @predicates[0].missing?(features)
+      end
+
       def or?(features)
-        @predicates.any? {|p| p.true?(features)}
+        @predicates.any? { |p| p.true?(features) }
       end
 
       def and?(features)
-        @predicates.all? {|p| p.true?(features)}
+        @predicates.all? { |p| p.true?(features) }
       end
     end
   end
-end
+end

data/lib/scoruby/predicates/false_predicate.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Scoruby
   module Predicates
     class FalsePredicate
@@ -9,9 +11,9 @@ module Scoruby
         false
       end
 
-      def is_missing?(_)
+      def missing?(_)
         false
       end
     end
   end
-end
+end

data/lib/scoruby/predicates/simple_predicate.rb

@@ -1,12 +1,16 @@
+# frozen_string_literal: true
+
 module Scoruby
   module Predicates
     class SimplePredicate
-
       GREATER_THAN     = 'greaterThan'
       LESS_THAN        = 'lessThan'
       LESS_OR_EQUAL    = 'lessOrEqual'
       GREATER_OR_EQUAL = 'greaterOrEqual'
-      MATH_OPS         = [GREATER_THAN, LESS_THAN, LESS_OR_EQUAL, GREATER_OR_EQUAL]
+      MATH_OPS         = [GREATER_THAN,
+                          LESS_THAN,
+                          LESS_OR_EQUAL,
+                          GREATER_OR_EQUAL].freeze
       EQUAL            = 'equal'
       IS_MISSING       = 'isMissing'
 
@@ -24,10 +28,10 @@ module Scoruby
       def true?(features)
         return num_true?(features) if MATH_OPS.include?(@operator)
         return features[@field] == @value if @operator == EQUAL
-        features[field].nil? || !features.has_key?(field) if @operator == IS_MISSING
+        features[field].nil? || !features.key?(field) if @operator == IS_MISSING
       end
 
-      def is_missing?(features)
+      def missing?(features)
         !features.keys.include?(@field)
       end
 
@@ -35,8 +39,10 @@ module Scoruby
 
       def num_true?(features)
         return false unless features[@field]
-        curr_value = Float(features[@field])
-        value      = Float(@value)
+        compare(Float(features[@field]), Float(@value))
+      end
+
+      def compare(curr_value, value)
         return curr_value > value if @operator == GREATER_THAN
         return curr_value < value if @operator == LESS_THAN
         return curr_value <= value if @operator == LESS_OR_EQUAL

data/lib/scoruby/predicates/simple_set_predicate.rb

@@ -1,7 +1,8 @@
+# frozen_string_literal: true
+
 module Scoruby
   module Predicates
     class SimpleSetPredicate
-
       IS_IN = 'isIn'
 
       attr_reader :field
@@ -17,16 +18,16 @@ module Scoruby
         @array.include? features[@field] if @operator == IS_IN
       end
 
-      def is_missing?(features)
+      def missing?(features)
         !features.keys.include?(@field)
       end
 
       private
 
       def single_or_quoted_words(string)
-        string.split(/\s(?=(?:[^"]|"[^"]*")*$)/).
-          reject(&:empty?).
-          map {|w| w.tr('"', '')}
+        string.split(/\s(?=(?:[^"]|"[^"]*")*$)/)
+              .reject(&:empty?)
+              .map { |w| w.tr('"', '') }
       end
     end
   end

data/lib/scoruby/predicates/true_predicate.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Scoruby
   module Predicates
     class TruePredicate
@@ -9,7 +11,7 @@ module Scoruby
         true
       end
 
-      def is_missing?(_)
+      def missing?(_)
         false
       end
     end

data/lib/scoruby/version.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 module Scoruby
-  VERSION = '0.2.7'
+  VERSION = '0.2.8'
 end

data/lib/scoruby.rb

@@ -1,3 +1,5 @@
+# frozen_string_literal: true
+
 require 'scoruby/version'
 require 'scoruby/models_factory'
 require 'nokogiri'
@@ -9,12 +11,12 @@ module Scoruby
 
     def logger
       @logger ||= Logger.new($stdout).tap do |log|
-        log.progname = self.name
+        log.progname = name
       end
     end
   end
 
-  def self.get_model(pmml_file_name)
+  def self.load_model(pmml_file_name)
     xml = xml_from_file_path(pmml_file_name)
     ModelsFactory.factory_for(xml)
   end
@@ -25,7 +27,7 @@ module Scoruby
   end
 
   def self.xml_from_string(pmml_string)
-    xml = Nokogiri::XML(pmml_string) { |config| config.noblanks }
+    xml = Nokogiri::XML(pmml_string, &:noblanks)
     xml.remove_namespaces!
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: scoruby
 version: !ruby/object:Gem::Version
-  version: 0.2.7
+  version: 0.2.8
 platform: ruby
 authors:
 - Asaf Schers
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2017-10-13 00:00:00.000000000 Z
+date: 2017-12-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler
@@ -117,6 +117,7 @@ extra_rdoc_files: []
 files:
 - ".gitignore"
 - ".rspec"
+- ".rubocop.yml"
 - ".travis.yml"
 - CODE_OF_CONDUCT.md
 - Gemfile
@@ -167,7 +168,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.4.8
+rubygems_version: 2.6.13
 signing_key: 
 specification_version: 4
 summary: Ruby Scoring API for PMML.