pest 0.0.0

data/lib/pest/estimator/frequency.rb

@@ -0,0 +1,61 @@
+require 'narray'
+
+class Pest::Estimator::Frequency
+  include Pest::Estimator
+  include Pest::Function::Probability
+  include Pest::Function::Entropy
+
+  def distribution_class
+    Distribution
+  end
+
+  class Distribution
+    include Pest::Estimator::Distribution
+
+    attr_reader :frequencies, :checksum
+
+    def cache_model
+      if @frequencies.nil?
+        @frequencies = Hash.new(0)
+        @estimator.data.data_vectors(variable_array).each do |vector|
+          # Make sure this vector is consistently ordered
+          @frequencies[vector] += 1
+        end
+      end
+    end
+
+    def probability(data)
+      cache_model
+
+      NArray[ data.data_vectors(variable_array).map do |vector|
+        @frequencies[vector].to_f
+      end ] / @estimator.data.length
+    end
+
+    def entropy
+      cache_model
+
+      probabilities = probability(unique_event_dataset)
+
+      (-probabilities * NMath.log2(probabilities)).sum
+    end
+
+    private
+
+    def unique_event_dataset
+      vectors = Pest::DataSet::NArray[@frequencies.keys]
+      hash = {}
+      variable_array.each_index do |i|
+        # Extract a single variable from the array of vectors
+        hash[variable_array[i]] = vectors[i,true,true].reshape!(vectors.shape[1]).to_a
+      end
+      Pest::DataSet::NArray.from_hash(hash)
+    end
+
+    def find_tempfile
+      if path = Dir.glob("#{Dir::Tmpname.tmpdir}/*").select {|path| path =~ /#{@checksum}\.#{@variables.hash}/}.first
+        File.open(path)
+      end
+    end
+  end
+end

data/lib/pest/function.rb

@@ -0,0 +1,28 @@
+module Pest::Function
+
+  module Builder
+    # Delegate methods to the result of 'evaluate'
+    #
+    methods = [
+      :sprintf, :+, :-, :*, :/, :**, :<=>, :==, :coerce, :floor, :ceil,
+      :truncate, :round, :to_i, :to_f, :to_r, :rationalize, :hash, :to_s,
+      :i, :+@, :-@, :eql?, :div, :divmod, :%, :modulo, :remainder, :abs,
+      :magnitude, :to_int, :real?, :integer?, :zero?, :nonzero?,
+      :>, :>=, :<, :<=, :between?, :pretty_print_instance_variables,
+      :pretty_print_inspect, :nil?, :===, :=~, :!~, :!, :!=
+    ]
+
+    methods.each do |f|
+      define_method(f) do |*args|
+        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

@@ -0,0 +1,35 @@
+module Pest::Function
+  module Entropy
+    def entropy(*variables)
+      Builder.new(self, variables)
+    end
+    alias :h :entropy
+
+    class Builder
+      include Pest::Function::Builder
+
+      attr_reader :estimator, :event, :givens
+
+      def initialize(estimator, variables)
+        @estimator      = estimator
+        @event          = parse(variables)
+        @givens         = [].to_set
+      end
+
+      def given(*variables)
+        @givens.merge parse(variables)
+        self
+      end
+
+      def evaluate
+        joint = estimator.distributions[event].entropy
+        if givens.empty?
+          joint
+        else
+          conditional = estimator.distributions[givens].entropy
+          joint - conditional
+        end
+      end
+    end
+  end
+end

data/lib/pest/function/probability.rb

@@ -0,0 +1,41 @@
+module Pest::Function
+  module Probability
+    def probability(*variables)
+      Builder.new(self, variables)
+    end
+    alias :p :probability
+
+    class Builder
+      include Pest::Function::Builder
+
+      attr_reader :estimator, :data_source, :event, :givens
+
+      def initialize(estimator, variables)
+        @estimator      = estimator
+        @data_source    = data_source
+        @event          = parse(variables)
+        @givens         = [].to_set
+      end
+
+      def given(*variables)
+        @givens.merge parse(variables)
+        self
+      end
+
+      def in(data_set)
+        @data_source = data_set
+        self
+      end
+
+      def evaluate
+        joint = estimator.distributions[event].probability(data_source)
+        if givens.empty?
+          joint
+        else
+          conditional = estimator.distributions[givens].probability(data_source)
+          joint / conditional
+        end
+      end
+    end
+  end
+end

data/lib/pest/variable.rb

@@ -0,0 +1,34 @@
+class Pest::Variable
+  def self.deserialize(string)
+    if string.chomp =~ /^([^\:]+)\:(\w{8}-\w{4}-\w{4}-\w{4}-\w{12})$/
+      new :name => $1, :uuid => $2
+    else
+      raise "Unable to parse string"
+    end
+  end
+
+  attr_reader :name, :uuid
+
+  def initialize(args={})
+    @name = args[:name]
+    @uuid = args[:uuid] || UUIDTools::UUID.random_create
+  end
+
+  def identifier
+    "#{name}:#{uuid}"
+  end
+  alias :serialize :identifier
+
+  def hash
+    identifier.hash
+  end
+
+  def ==(other)
+    other.kind_of?(self.class) and identifier == other.identifier
+  end
+  alias :eql? :==
+
+  def <=>(other)
+    identifier <=> other.identifier
+  end
+end

data/lib/pest/version.rb

@@ -0,0 +1,7 @@
+module Pest
+  VERSION = "0.0.1"
+
+  def self.version
+    VERSION
+  end
+end

data/pest.gemspec

@@ -0,0 +1,112 @@
+# Generated by jeweler
+# DO NOT EDIT THIS FILE DIRECTLY
+# Instead, edit Jeweler::Tasks in Rakefile, and run 'rake gemspec'
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = "pest"
+  s.version = "0.0.0"
+
+  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.description = "Wrappers to facilitate different classes of probability estimators"
+  s.email = "kerinin@gmail.com"
+  s.extra_rdoc_files = [
+    "README.md"
+  ]
+  s.files = [
+    ".travis.yml",
+    "Gemfile",
+    "README.md",
+    "Rakefile",
+    "VERSION",
+    "lib/pest.rb",
+    "lib/pest/data_set.rb",
+    "lib/pest/data_set/hash.rb",
+    "lib/pest/data_set/narray.rb",
+    "lib/pest/estimator.rb",
+    "lib/pest/estimator/frequency.rb",
+    "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",
+    "spec/pest/data_set/narray_spec.rb",
+    "spec/pest/data_set_spec.rb",
+    "spec/pest/estimator/bernoulli_spec.rb",
+    "spec/pest/estimator/frequency_spec.rb",
+    "spec/pest/estimator/gaussian_spec.rb",
+    "spec/pest/estimator/multinomial_spec.rb",
+    "spec/pest/estimator/parzen_spec.rb",
+    "spec/pest/estimator/svd_spec.rb",
+    "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.summary = "Probability Estimation"
+
+  if s.respond_to? :specification_version then
+    s.specification_version = 3
+
+    if Gem::Version.new(Gem::VERSION) >= Gem::Version.new('1.2.0') then
+      s.add_runtime_dependency(%q<pest>, [">= 0"])
+      s.add_runtime_dependency(%q<narray>, [">= 0"])
+      s.add_runtime_dependency(%q<uuidtools>, [">= 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"])
+      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"])
+      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"])
+      s.add_dependency(%q<uuidtools>, [">= 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"])
+      s.add_dependency(%q<jeweler>, [">= 0"])
+      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"])
+    s.add_dependency(%q<narray>, [">= 0"])
+    s.add_dependency(%q<uuidtools>, [">= 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"])
+    s.add_dependency(%q<jeweler>, [">= 0"])
+    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

@@ -0,0 +1,108 @@
+require 'spec_helper'
+
+# NOTE: make sure you're clear on the relationship between variables,
+# their names, and the keys used in @hash.  I think they're getting
+# 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
+
+      it "maps Symbol => from_file" do
+        @class.translators[Symbol].should == :from_file
+      end
+    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 "generates variables" do
+        @class.from_file('foo').variables[:foo].should be_a(Pest::Variable)
+      end
+    end
+  end
+
+  before(:each) do
+    @data = {:foo => [1,2,3], :bar => [3,4,5]}
+    @instance = Pest::DataSet::Hash.new(@data)
+  end
+
+  describe "#to_hash" do
+    it "returns a hash" do
+      @instance.to_hash.should == @data
+    end
+  end
+
+  describe "#data_vectors" do
+    it "returns an enumerable" do
+      @instance.data_vectors.should be_a(Enumerable)
+    end
+
+    it "formats data as a list of rows" do
+      @instance.data_vectors.first.should == [1,3]
+    end
+  end
+
+  describe "#save" do
+    before(:each) do
+      @file = File.open(__FILE__, 'r')
+      File.stub(:open).with('foo', 'w').and_return(@file)
+      Marshal.stub(:dump)
+    end
+
+    it "marshals to file from path" do
+      Marshal.should_receive(:dump).with(@data, @file)
+      @instance.save('foo')
+    end
+
+    it "marshals to file from file" do
+      Marshal.should_receive(:dump).with(@data, @file)
+      @instance.save(@file)
+    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
+    end
+  end
+
+  describe "#length" do
+    before(:each) do
+      @data = {:foo => [1,2,3], :bar => [3,4,5]}
+      @instance = Pest::DataSet::Hash.new(@data)
+    end
+
+    it "delegates to hash" do
+      @instance.length.should == 3
+    end
+  end
+end

data/spec/pest/data_set/narray_spec.rb

@@ -0,0 +1,141 @@
+require 'spec_helper'
+require 'narray'
+
+describe Pest::DataSet::NArray do
+  before(:each) do
+    @v1 = Pest::Variable.new(:name => :foo)
+    @v2 = Pest::Variable.new(:name => :bar)
+    @class = Pest::DataSet::NArray
+  end
+
+  describe "::translators" do
+    it "maps Pest::DataSet::Hash => from_hash" do
+      @class.translators[Pest::DataSet::Hash].should == :from_hash
+    end
+
+    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
+  end
+
+  describe "::from_file" do
+    before(:each) do
+      @matrix = @class.from_hash @v1 => [1,2,3], @v2 => [4,5,6]
+      @file = Tempfile.new('test')
+      @matrix.save(@file.path)
+    end
+
+    it "loads from NArray IO" do
+      @class.from_file(@file).should == @matrix
+    end
+
+    it "looks for NArray from string" do
+      @class.from_file(@file.path).should == @matrix
+    end
+      
+    it "sets variables" do
+      @class.from_file(@file).variables.values.should == [@v1, @v2]
+    end
+  end
+
+  describe "::from_hash" do
+    before(:each) do
+      @matrix = NArray.to_na [[4,5,6],[1,2,3]]
+    end
+
+    it "creates a NArray" do
+      @class.from_hash({@v1 => [1,2,3], @v2 => [4,5,6]}).should == @matrix
+    end
+
+    it "sets variables" do
+      @class.from_hash({@v1 => [1,2,3], @v2 => [4,5,6]}).variables.values.should == [@v1, @v2]
+    end
+
+    it "generates Pest::Variables if not passed" do
+      @class.from_hash({:foo => [1,2,3]}).variables[:foo].should be_a(Pest::Variable)
+    end
+  end
+
+  describe "::from_csv" do
+    before(:each) do
+      @file = Tempfile.new('test_csv')
+      CSV.open(@file.path, 'w') do |csv|
+        csv << ["foo", "bar"]
+        csv << [1,1]
+        csv << [1,2]
+        csv << [1,3]
+      end
+    end
+
+    it "creates variables from first line" do
+      @instance = @class.from_csv @file.path
+      @instance.variable_array.map(&:name).should == ["bar", "foo"]
+    end
+
+    it "creates data from the rest" do
+      @instance = @class.from_csv @file.path
+      @instance.to_hash.should == {@instance.variables["foo"] => [1,1,1], @instance.variables["bar"] => [1,2,3]}
+    end
+
+    it "accepts a filename" do
+      @instance = @class.from_csv @file.path
+      @instance.to_hash.should == {@instance.variables["foo"] => [1,1,1], @instance.variables["bar"] => [1,2,3]}
+    end
+
+    it "accepts an IO" do
+      @instance = @class.from_csv @file
+      @instance.to_hash.should == {@instance.variables["foo"] => [1,1,1], @instance.variables["bar"] => [1,2,3]}
+    end
+  end
+
+  describe "#to_hash" do
+    before(:each) do
+      @instance = @class.from_hash :foo => [1,2,3], :bar => [4,5,6]
+    end
+
+    it "sets keys" do
+      @instance.to_hash.keys.should == @instance.variables.values
+    end
+
+    it "sets values" do
+      @instance.to_hash.values.should == [[4,5,6],[1,2,3]]
+    end
+  end
+
+  describe "#data_vectors" do
+    before(:each) do
+      @instance = @class.from_hash :foo => [1,2,3], :bar => [4,5,6]
+    end
+
+    it "returns an enumerable" do
+      @instance.data_vectors.should be_a(Enumerable)
+    end
+
+    it "slices" do
+      # NOTE: This is returning an array - probably could be more efficient
+      @instance.data_vectors.first.should == [4,1]
+    end
+  end
+
+  describe "#save" do
+    before(:each) do
+      @file = Tempfile.new('test')
+      @instance = @class.from_hash :foo => [1,2,3], :bar => [4,5,6]
+    end
+
+    it "marshals to file" do
+      @instance.save(@file)
+      @class.from_file(@file.path).should == @instance
+    end
+
+    it "saves to tmp dir if no filename specified" do
+      Tempfile.should_receive(:new).and_return(@file)
+      @instance.save
+      @class.from_file(@file.path).should == @instance
+    end
+  end
+end