pest 0.0.0 → 0.1.0

data/lib/pest/estimator.rb

@@ -13,17 +13,6 @@ module Pest::Estimator
     @distributions ||= DistributionList.new(self)
   end
 
-  def to_variable(arg)
-    variable = case arg.class.name
-    when 'Pest::Variable'
-      arg
-    when 'String', 'Symbol'
-      variables[arg] || Pest::Variable.new(:name => arg)
-    end
-    raise ArgumentError unless variables.values.include?(variable)
-    variable
-  end
-
   module Distribution
     attr_reader :variables
 
@@ -36,7 +25,11 @@ module Pest::Estimator
       variables.to_a.sort
     end
 
-    def probability
+    def batch_probability(*args)
+      raise NotImplementedError
+    end
+
+    def probability(*args)
       raise NotImplementedError
     end
   end
@@ -46,23 +39,10 @@ module Pest::Estimator
       @estimator = estimator
     end
 
-    def parse_args(args)
-      set = if args.kind_of? Array
-        if args.any? {|arg| arg.kind_of?(::Set)}
-          args.inject(::Set.new) {|set, el| set + el.to_set}
-        else
-          args.flatten.to_set
-        end
-      elsif args.kind_of? ::Set
-        args
-      else
-        Array(args).to_set
-      end
-      set.map! {|arg| @estimator.to_variable(arg) }
-    end
-
     def [](*args)
-      set = parse_args(args)
+      set = args.to_set
+      raise ArgumentError unless (set - @estimator.variables).empty?
+
       unless has_key? set
         self[set] = @estimator.distribution_class.new(@estimator, set)
       end

data/lib/pest/estimator/frequency.rb

@@ -11,15 +11,16 @@ class Pest::Estimator::Frequency
 
   class Distribution
     include Pest::Estimator::Distribution
-
+    
+    OFFSET = 0
     attr_reader :frequencies, :checksum
 
     def cache_model
       if @frequencies.nil?
         @frequencies = Hash.new(0)
-        @estimator.data.data_vectors(variable_array).each do |vector|
+        @estimator.data.pick(*variable_array).each do |vector|
           # Make sure this vector is consistently ordered
-          @frequencies[vector] += 1
+          @frequencies[Array(vector)] += 1
         end
       end
     end
@@ -27,9 +28,10 @@ class Pest::Estimator::Frequency
     def probability(data)
       cache_model
 
-      NArray[ data.data_vectors(variable_array).map do |vector|
-        @frequencies[vector].to_f
-      end ] / @estimator.data.length
+      array = NArray[ data.pick(*variable_array).map do |vector|
+        @frequencies[Array(vector)].to_f
+      end ]
+      (OFFSET + array.reshape!(data.length)) / (OFFSET + @estimator.data.length)
     end
 
     def entropy
@@ -43,7 +45,7 @@ class Pest::Estimator::Frequency
     private
 
     def unique_event_dataset
-      vectors = Pest::DataSet::NArray[@frequencies.keys]
+      vectors = NMatrix[@frequencies.keys]
       hash = {}
       variable_array.each_index do |i|
         # Extract a single variable from the array of vectors

data/lib/pest/function.rb

@@ -17,12 +17,5 @@ module Pest::Function
         evaluate.send(f, *args)
       end
     end
-
-    private
-
-    def parse(variables)
-      variables.map {|arg| estimator.to_variable(arg) }.to_set
-    end
-
   end
 end

data/lib/pest/function/entropy.rb

@@ -12,12 +12,14 @@ module Pest::Function
 
       def initialize(estimator, variables)
         @estimator      = estimator
-        @event          = parse(variables)
-        @givens         = [].to_set
+        @event          = variables.to_set
+        @givens         = Set.new
+        raise ArgumentError unless (@event - @estimator.variables).empty?
       end
 
       def given(*variables)
-        @givens.merge parse(variables)
+        @givens.merge variables.to_set
+        raise ArgumentError unless (@givens - @estimator.variables).empty?
         self
       end
 

data/lib/pest/function/probability.rb

@@ -1,24 +1,30 @@
 module Pest::Function
   module Probability
-    def probability(*variables)
-      Builder.new(self, variables)
+    def batch_probability(*variables)
+      BatchBuilder.new(self, variables)
+    end
+    alias :batch_p :batch_probability
+
+    def probability(event={})
+      Builder.new(self, event)
     end
     alias :p :probability
 
-    class Builder
+    class BatchBuilder
       include Pest::Function::Builder
 
       attr_reader :estimator, :data_source, :event, :givens
 
       def initialize(estimator, variables)
         @estimator      = estimator
-        @data_source    = data_source
-        @event          = parse(variables)
+        @event          = variables.to_set
         @givens         = [].to_set
+        raise ArgumentError unless (@event - @estimator.variables).empty?
       end
 
       def given(*variables)
-        @givens.merge parse(variables)
+        @givens += variables
+        raise ArgumentError unless (@givens - @estimator.variables).empty?
         self
       end
 
@@ -28,14 +34,43 @@ module Pest::Function
       end
 
       def evaluate
-        joint = estimator.distributions[event].probability(data_source)
         if givens.empty?
-          joint
+          estimator.distributions[*event].probability(data_source).to_a
         else
-          conditional = estimator.distributions[givens].probability(data_source)
-          joint / conditional
+          joint = estimator.distributions[*(event + givens)].probability(data_source)
+          conditional = estimator.distributions[*givens].probability(data_source)
+
+          (joint / conditional).to_a
         end
       end
     end
+
+    class Builder
+      include Pest::Function::Builder
+
+      attr_accessor :estimator, :event, :givens
+
+      def initialize(estimator, event)
+        @estimator    = estimator
+        @event        = event
+        @givens       = Hash.new
+      end
+
+      def given(given)
+        givens.merge! given
+        raise ArgumentError unless (given.keys.to_set - @estimator.variables).empty?
+        self
+      end
+
+      def evaluate
+        data_hash = event.merge(givens)
+        data_hash.each_key {|key| data_hash[key] = Array(data_hash[key])}
+
+        data = Pest::DataSet::Hash.from_hash(data_hash)
+        BatchBuilder.new(estimator, event.keys).
+          given(*givens.keys).in(data).
+          evaluate.first
+      end
+    end
   end
 end

data/pest.gemspec

@@ -5,11 +5,11 @@
 
 Gem::Specification.new do |s|
   s.name = "pest"
-  s.version = "0.0.0"
+  s.version = "0.1.0"
 
   s.required_rubygems_version = Gem::Requirement.new(">= 0") if s.respond_to? :required_rubygems_version=
   s.authors = ["Ryan Michael"]
-  s.date = "2012-06-24"
+  s.date = "2012-07-05"
   s.description = "Wrappers to facilitate different classes of probability estimators"
   s.email = "kerinin@gmail.com"
   s.extra_rdoc_files = [
@@ -30,7 +30,6 @@ Gem::Specification.new do |s|
     "lib/pest/function.rb",
     "lib/pest/function/entropy.rb",
     "lib/pest/function/probability.rb",
-    "lib/pest/variable.rb",
     "lib/pest/version.rb",
     "pest.gemspec",
     "spec/pest/data_set/hash_spec.rb",
@@ -45,14 +44,13 @@ Gem::Specification.new do |s|
     "spec/pest/estimator_spec.rb",
     "spec/pest/function/entropy_spec.rb",
     "spec/pest/function/probability_spec.rb",
-    "spec/pest/variable_spec.rb",
     "spec/pest_spec.rb",
     "spec/spec_helper.rb"
   ]
   s.homepage = "http://github.com/kerinin/pest"
   s.licenses = ["MIT"]
   s.require_paths = ["lib"]
-  s.rubygems_version = "1.8.24"
+  s.rubygems_version = "1.8.23"
   s.summary = "Probability Estimation"
 
   if s.respond_to? :specification_version then
@@ -74,6 +72,10 @@ Gem::Specification.new do |s|
       s.add_development_dependency(%q<rake>, [">= 0"])
       s.add_development_dependency(%q<pry>, [">= 0"])
       s.add_development_dependency(%q<rspec>, [">= 0"])
+      s.add_development_dependency(%q<jeweler>, [">= 0"])
+      s.add_development_dependency(%q<rake>, [">= 0"])
+      s.add_development_dependency(%q<pry>, [">= 0"])
+      s.add_development_dependency(%q<rspec>, [">= 0"])
     else
       s.add_dependency(%q<pest>, [">= 0"])
       s.add_dependency(%q<narray>, [">= 0"])
@@ -90,6 +92,10 @@ Gem::Specification.new do |s|
       s.add_dependency(%q<rake>, [">= 0"])
       s.add_dependency(%q<pry>, [">= 0"])
       s.add_dependency(%q<rspec>, [">= 0"])
+      s.add_dependency(%q<jeweler>, [">= 0"])
+      s.add_dependency(%q<rake>, [">= 0"])
+      s.add_dependency(%q<pry>, [">= 0"])
+      s.add_dependency(%q<rspec>, [">= 0"])
     end
   else
     s.add_dependency(%q<pest>, [">= 0"])
@@ -107,6 +113,10 @@ Gem::Specification.new do |s|
     s.add_dependency(%q<rake>, [">= 0"])
     s.add_dependency(%q<pry>, [">= 0"])
     s.add_dependency(%q<rspec>, [">= 0"])
+    s.add_dependency(%q<jeweler>, [">= 0"])
+    s.add_dependency(%q<rake>, [">= 0"])
+    s.add_dependency(%q<pry>, [">= 0"])
+    s.add_dependency(%q<rspec>, [">= 0"])
   end
 end
 

data/spec/pest/data_set/hash_spec.rb

@@ -5,55 +5,30 @@ require 'spec_helper'
 # conflated
 
 describe Pest::DataSet::Hash do
-  context "class methods" do
-    before(:each) do
-      @class = Pest::DataSet::Hash
-    end
-
-    describe "::translators" do
-      it "maps File => from_file" do
-        @class.translators[File].should == :from_file
-      end
-
-      it "maps String => from_file" do
-        @class.translators[String].should == :from_file
-      end
+  before(:each) do
+    @class = Pest::DataSet::Hash
+  end
 
-      it "maps Symbol => from_file" do
-        @class.translators[Symbol].should == :from_file
-      end
+  describe "::translators" do
+    it "maps String => from_file" do
+      @class.translators[String].should == :from_file
     end
 
-    describe "::from_file" do
-      before(:each) do
-        file = File.open(__FILE__, 'r')
-        File.stub(:open).with('foo', 'r').and_return(file)
-        Marshal.stub(:restore).with(file).and_return({:foo => 1})
-      end
-
-      it "looks for file if passed string" do
-        File.should_receive(:open).with('foo', 'r')
-        @class.from_file('foo')
-      end
-
-      it "unmarshals" do
-        Marshal.should_receive(:restore)
-        @class.from_file('foo')
-      end
-
-      it "sets variables" do
-        @class.from_file('foo').variables.length.should == 1
-      end
+    it "maps Symbol => from_file" do
+      @class.translators[Symbol].should == :from_file
+    end
+  end
 
-      it "generates variables" do
-        @class.from_file('foo').variables[:foo].should be_a(Pest::Variable)
-      end
+  describe "::from_hash" do
+    it "gets variables from keys" do
+      @instance = Pest::DataSet::Hash.from_hash(:foo => [1,2,3], :bar => [3,4,5])
+      @instance.variables.should == [:foo, :bar].to_set
     end
   end
 
   before(:each) do
     @data = {:foo => [1,2,3], :bar => [3,4,5]}
-    @instance = Pest::DataSet::Hash.new(@data)
+    @instance = Pest::DataSet::Hash.from_hash(@data)
   end
 
   describe "#to_hash" do
@@ -62,47 +37,87 @@ describe Pest::DataSet::Hash do
     end
   end
 
-  describe "#data_vectors" do
-    it "returns an enumerable" do
-      @instance.data_vectors.should be_a(Enumerable)
+  describe "#pick" do
+    before(:each) do
+      @instance = @class.from_hash :foo => [1,2,3], :bar => [4,5,6]
+    end
+
+    it "accepts a single symbol string" do
+      @instance.pick(:foo).data.to_a.first.should == [1,2,3]
     end
 
-    it "formats data as a list of rows" do
-      @instance.data_vectors.first.should == [1,3]
+    it "accepts a single variable" do
+      @instance.pick(:foo).data.to_a.first.should == [1,2,3]
     end
+
+    it "accepts multiple variables" do
+      @instance.pick(:bar, :foo).data.to_a.should == [[4,5,6],[1,2,3]]
+   end
   end
 
-  describe "#save" do
+  describe "#length" do
     before(:each) do
-      @file = File.open(__FILE__, 'r')
-      File.stub(:open).with('foo', 'w').and_return(@file)
-      Marshal.stub(:dump)
+      @data = {:foo => [1,2,3], :bar => [3,4,5]}
+      @instance = Pest::DataSet::Hash.from_hash(@data)
     end
 
-    it "marshals to file from path" do
-      Marshal.should_receive(:dump).with(@data, @file)
-      @instance.save('foo')
+    it "delegates to hash" do
+      @instance.length.should == 3
     end
+  end
 
-    it "marshals to file from file" do
-      Marshal.should_receive(:dump).with(@data, @file)
-      @instance.save(@file)
+  describe "#each" do
+    before(:each) do
+      @instance = @class.from_hash :foo => [1,2,3], :bar => [4,5,6]
     end
 
-    it "saves to tmp dir if no filename specified" do
-      File.should_receive(:open).with(/pest_hash_dataset/, anything(), anything()).and_return(@file)
-      @instance.save
+    it "yields vectors" do
+      block = double("block")
+      block.should_receive(:yielding).with([1,4])
+      block.should_receive(:yielding).with([2,5])
+      block.should_receive(:yielding).with([3,6])
+      @instance.each {|i| block.yielding(i)}
     end
   end
 
-  describe "#length" do
+  describe "#map" do
     before(:each) do
-      @data = {:foo => [1,2,3], :bar => [3,4,5]}
-      @instance = Pest::DataSet::Hash.new(@data)
+      @instance = @class.from_hash :foo => [1,2,3], :bar => [4,5,6]
     end
 
-    it "delegates to hash" do
-      @instance.length.should == 3
+    it "works" do
+      @instance.map {|i| i}.should == [[1,4],[2,5],[3,6]]
+    end
+  end
+
+  describe "#merge" do
+    before(:each) do
+      @other    = @class.from_hash :foo => [10,11,12,13], :baz => [1,2,3,4]
+      @instance = @class.from_hash :foo => [1,2,3,4], :bar => [5,6,7,8]
+    end
+
+    it "accepts a dataset and returns dataset" do
+      @instance.merge(@other).should be_a(@class)
+    end
+
+    it "accepts a hash and returns dataset" do
+      @instance.merge(:foo => [10,11,12,13], :baz => [1,2,3,4]).should be_a(@class)
+    end
+
+    it "requires the dataset to have the same length" do
+      expect { @instance.merge(:foo => [1,2,3,4,5]) }.to raise_error(ArgumentError)
+    end
+
+    it "adds the passed variable to self" do
+      @instance.merge(@other).variables.should include(:baz)
+    end
+
+    it "adds the passed data to self" do
+      @instance.merge(@other).pick(:baz).to_a.flatten.should == [1,2,3,4]
+    end
+
+    it "over-writes variables in self with variables in other" do
+      @instance.merge(@other).pick(:foo).to_a.flatten.should == [10,11,12,13]
     end
   end
 end