fathom 0.3.0 → 0.3.1

data/spec/fathom/node/cpm_node_spec.rb

@@ -0,0 +1,144 @@
+require File.expand_path('../../../spec_helper', __FILE__)
+
+describe CPMNode do
+  
+  before do
+    @child = BeliefNode.new(:name => :color, :values => {:red => 0.4, :green => 0.6})
+    @parent = BeliefNode.new(:name => :child, :values => {:asher => 0.2, :stella => 0.8})
+    @cpm = CPMNode.new(:child => @child, :parent => @parent)
+  end
+  
+  it "should subclass Node" do
+    CPMNode.ancestors[1].should eql(Node)
+  end
+  
+  it "should allow a BeliefNode child and parent" do
+    @cpm.parent.should eql(@parent)
+    @cpm.child.should eql(@child)
+  end
+  
+  it "should generate a conditional probability matrix from the parent and child" do
+    parent_values = @cpm.parent.probabilities.col
+    child_values = @cpm.child.probabilities
+    matrix = parent_values * child_values
+    @cpm.values.to_a.should eql(matrix.to_a)
+  end
+  
+  it "should alias the values with matrix" do
+    @cpm.values.should eql(@cpm.matrix)
+  end
+  
+  it "should raise an error when trying to load non-BeliefNode parent or child" do
+    lambda{CPMNode.new(:child => :not_a_belief_node, :parent => @parent)}.should raise_error(/must be a BeliefNode/)
+    lambda{CPMNode.new(:child => @child, :parent => :not_a_belief_node)}.should raise_error(/must be a BeliefNode/)
+  end
+  
+  it "should have a probability lookup" do
+    @cpm.should be_respond_to(:probability)
+  end
+
+  it "should only allow options for the parent and child names" do
+    lambda{@cpm.probability :not_a_link => true}.should raise_error(/unknown/i)
+  end
+
+  it "should be able to lookup the values we're looking for" do
+    # The Matrix looks like this:
+    # GSL::Matrix
+    # [  8.000e-02  1.200e-01 
+    #    3.200e-01  4.800e-01 ]
+    @cpm.probability(:color => :red, :child => :asher).should be_close(0.08, 1e-10)
+    @cpm.probability(:color => :red, :child => :stella).should be_close(0.32, 1e-10)
+    @cpm.probability(:color => :green, :child => :asher).should be_close(0.12, 1e-10)
+    @cpm.probability(:color => :green, :child => :stella).should be_close(0.48, 1e-10)
+  end
+  
+  it "should look for all children values, if no child is provided" do
+    @cpm.probability(:child => :asher).should be_close(0.2, 1e-10)
+    @cpm.probability(:child => :stella).should be_close(0.8, 1e-10)
+  end
+  
+  it "should look for all parent values, if no parent is provided" do
+    @cpm.probability(:color => :red).should be_close(0.4, 1e-10)
+    @cpm.probability(:color => :green).should be_close(0.6, 1e-10)
+  end
+  
+  it "should provide a probability of 1 if no filter is set" do
+    @cpm.probability.should be_close(1.0, 1e-10)
+  end
+  
+  it "should give a description in a hash, if description is turned on" do
+    @cpm.probability(:color => :red, :child => :asher, :describe => true).keys.first.should eql('P(red | asher)')
+    @cpm.probability(:color => :red, :child => :stella, :describe => true).keys.first.should eql('P(red | stella)')
+    @cpm.probability(:color => :green, :child => :asher, :describe => true).keys.first.should eql('P(green | asher)')
+    @cpm.probability(:color => :green, :child => :stella, :describe => true).keys.first.should eql('P(green | stella)')
+    @cpm.probability(:child => :asher, :describe => true).keys.first.should eql('P(red or green | asher)')
+    @cpm.probability(:child => :stella, :describe => true).keys.first.should eql('P(red or green | stella)')
+    @cpm.probability(:color => :red, :describe => true).keys.first.should eql('P(red | asher or stella)')
+    @cpm.probability(:color => :green, :describe => true).keys.first.should eql('P(green | asher or stella)')
+    @cpm.probability(:describe => true).keys.first.should eql('P(red or green | asher or stella)')
+  end
+  
+  it "should alias p for probability" do
+    @cpm.p(:color => :red, :child => :asher).should be_close(0.08, 1e-10)
+    @cpm.p(:color => :red, :child => :stella).should be_close(0.32, 1e-10)
+    @cpm.p(:color => :green, :child => :asher).should be_close(0.12, 1e-10)
+    @cpm.p(:color => :green, :child => :stella).should be_close(0.48, 1e-10)
+    @cpm.p(:child => :asher).should be_close(0.2, 1e-10)
+    @cpm.p(:child => :stella).should be_close(0.8, 1e-10)
+    @cpm.p(:color => :red).should be_close(0.4, 1e-10)
+    @cpm.p(:color => :green).should be_close(0.6, 1e-10)
+    @cpm.p.should be_close(1.0, 1e-10)
+  end
+  
+  it "should have odds for any query" do
+    @cpm.odds(:color => :red, :child => :asher).should be_close(0.087, 1e-3)
+    @cpm.odds(:color => :red, :child => :stella).should be_close(0.471, 1e-3)
+    @cpm.odds(:color => :green, :child => :asher).should be_close(0.136, 1e-3)
+    @cpm.odds(:color => :green, :child => :stella).should be_close(0.923, 1e-3)
+    @cpm.odds(:child => :asher).should be_close(0.25, 1e-3)
+    @cpm.odds(:child => :stella).should be_close(4.0, 1e-3)
+    @cpm.odds(:color => :red).should be_close(0.666, 1e-3)
+    @cpm.odds(:color => :green).should be_close(1.5, 1e-3)
+    @cpm.odds.should eql(1 / 0.0)
+  end
+  
+  it "should alias o for odds" do
+    @cpm.o(:color => :red, :child => :asher).should be_close(0.087, 1e-3)
+    @cpm.o(:color => :red, :child => :stella).should be_close(0.471, 1e-3)
+    @cpm.o(:color => :green, :child => :asher).should be_close(0.136, 1e-3)
+    @cpm.o(:color => :green, :child => :stella).should be_close(0.923, 1e-3)
+    @cpm.o(:child => :asher).should be_close(0.25, 1e-3)
+    @cpm.o(:child => :stella).should be_close(4.0, 1e-3)
+    @cpm.o(:color => :red).should be_close(0.666, 1e-3)
+    @cpm.o(:color => :green).should be_close(1.5, 1e-3)
+    @cpm.o.should eql(1 / 0.0)
+  end
+  
+  it "should default the name to :cpm" do
+    @cpm.name.should eql(:cpm)
+  end
+  
+  it "should default the description" do
+    @cpm.description.should eql("Conditional Probability Matrix from child to color.")
+  end
+  
+  it "should return a vector for likelihood" do
+    @cpm.likelihood(:red).should be_a(GSL::Vector)
+  end
+  
+  it "should offer the likelihood of each parent value, given a child value" do
+    values = @cpm.likelihood(:red).to_a
+    values.sort.map {|e| (e * 100).to_i}.should eql([8, 32])
+
+    values = @cpm.likelihood(:green).to_a
+    values.sort.map {|e| (e * 100).to_i}.should eql([12, 48])
+  end
+  
+  it "should offer l as an alias for likelihood" do
+    values = @cpm.l(:red).to_a
+    values.sort.map {|e| (e * 100).to_i}.should eql([8, 32])
+
+    values = @cpm.l(:green).to_a
+    values.sort.map {|e| (e * 100).to_i}.should eql([12, 48])
+  end
+end

data/spec/fathom/node/data_collection_spec.rb

@@ -0,0 +1,26 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+include Fathom
+
+describe DataCollection do
+  it "should be a node" do
+    DataCollection.ancestors.should be_include(Node)
+  end
+  
+  it "should be a DiscreteNode as well" do
+    DataCollection.ancestors.should be_include(DiscreteNode)
+  end
+  
+  it "should be able to extract the discrete labels from the first discrete parent node added to it" do
+    pr = PlausibleRange.new(:min => 0, :max => 1)
+    dn = DiscreteNode.new(:labels => [1,2,3])
+    dc = DataCollection.new(:parents => [pr, dn])
+    dc.labels.should eql([1,2,3])
+  end
+  
+  it "should enforce a name" do
+    dc = DataCollection.new(:labels => [1,2,3])
+    dc.name.should_not be_nil
+  end
+  
+end

data/spec/fathom/{data_node_spec.rb → node/data_node_spec.rb}

@@ -1,4 +1,4 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
 
 include Fathom
 

data/spec/fathom/node/decision_spec.rb

@@ -0,0 +1,15 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+include Fathom
+
+describe Decision do
+  it "should be a type of node" do
+    Decision.ancestors.should be_include(Node)
+  end
+  
+  it "should not have a distribution or values" do
+    @d = Decision.new
+    @d.should_not be_respond_to(:distribution)
+    @d.should_not be_respond_to(:values)
+  end
+end

data/spec/fathom/node/discrete_node_spec.rb

@@ -0,0 +1,56 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+include Fathom
+
+describe DiscreteNode do
+  
+  before do
+    @dn = DiscreteNode.new(:labels => [1,2,3])
+  end
+  
+  it "should assign :true and :false for labels, if none are provided" do
+    lambda{@dn = DiscreteNode.new}.should_not raise_error
+    @dn.labels.should eql([:true, :false])
+  end
+  
+  it "should create an array from a single value for its labels" do
+    dn = DiscreteNode.new(:labels => 1)
+    dn.labels.should eql([1])
+  end
+  
+  it "should take the values and convert them to labels, if there are no labels provided" do
+    dn = DiscreteNode.new(:values => [1,2,3])
+    dn.labels.should eql([1,2,3])
+  end
+  
+  it "should create a unique list of labels" do
+    dn = DiscreteNode.new(:labels => [1,1,2,3,1])
+    dn.labels.should eql([1,2,3])
+  end
+  
+  it "should expose the size of the labels" do
+    @dn.size.should eql(3)
+  end
+  
+  it "should alias length for size" do
+    @dn.length.should eql(3)
+  end
+  
+  it "should have a default distribution of DiscreteUniform" do
+    @dn.distribution.should eql(Fathom::Distributions::DiscreteUniform)
+  end
+  
+  it "should be able to produce a random variable from the discrete distribution" do
+    @dn = DiscreteNode.new(:labels => [1,2])
+    # 9.31322574615479e-10 chance of failing
+    30.times.map {@dn.rand}.uniq.sort.should eql([1,2])
+  end
+  
+  it "should offer a uniform likelihood" do
+    @dn.likelihood(:anything).to_a.should eql([1.0, 1.0, 1.0])
+  end
+  
+  it "should offer an alias of l for likelihood" do
+    @dn.l(:anything).to_a.should eql([1.0, 1.0, 1.0])
+  end
+end

data/spec/fathom/node/fact_spec.rb

@@ -0,0 +1,33 @@
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
+
+include Fathom
+
+describe Fact do
+  before do
+    @f = Fact.new(
+      :value => 2, 
+      :name => "Some Fact", 
+      :description => "Just a demonstration of what a good fact may look like.  Facts are really about things that are certain--there are 7 days in a week, that sort of thing.  Too often I was creating a PlausibleRange with the min and max to the same value."
+    )
+  end
+  
+  it "should be a Node" do
+    Fact.ancestors.should be_include(Fathom::Node)
+  end
+  
+  it "should take a value (rather than values)" do
+    @f.value.should eql(2)
+  end
+  
+  it "should return the value for rand" do
+    @f.rand.should eql(2)
+  end
+  
+  it "should have removed the distribution and values methods" do
+    @f.should_not be_respond_to(:values)
+    @f.should_not be_respond_to(:distribution)
+    @n = Node.new(:distribution => :standard, :values => [1,2,3])
+    @n.values.should eql([1,2,3])
+    @n.distribution.should eql(Fathom::Distributions::Gaussian)
+  end
+end

data/spec/fathom/{mc_node_spec.rb → node/mc_node_spec.rb}

@@ -1,4 +1,4 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
 
 include Fathom
 

data/spec/fathom/{enforced_name_spec.rb → node/node_extensions/enforced_name_spec.rb}

@@ -1,4 +1,4 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
+require File.expand_path(File.dirname(__FILE__) + '/../../../spec_helper')
 
 include Fathom::Distributions
 

data/spec/fathom/{numeric_methods_spec.rb → node/node_extensions/numeric_methods_spec.rb}

@@ -1,4 +1,4 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
+require File.expand_path(File.dirname(__FILE__) + '/../../../spec_helper')
 
 include Fathom
 
@@ -11,11 +11,11 @@ describe NumericMethods do
   end
   
   before do
-    @n = DummyNumericNode.new(@opts)
+    @n = Distributions::DummyNumericNode.new(@opts)
   end
   
   it "should respond to a rand method" do
-    @n = DummyNumericNode.new
+    @n = Distributions::DummyNumericNode.new
     @n.should be_respond_to(:rand)
   end
   
@@ -33,7 +33,7 @@ describe NumericMethods do
   end
   
   it "should not raise an error when looking for a standard deviation and there is no vector" do
-    @n = DummyNumericNode.new
+    @n = Distributions::DummyNumericNode.new
     @n.vector.should be_nil
     @n.sd.should be_nil
   end
@@ -43,17 +43,17 @@ describe NumericMethods do
   end
   
   it "should return nil for mean if it doesn't have a vector" do
-    @n = DummyNumericNode.new
+    @n = Distributions::DummyNumericNode.new
     @n.mean.should be_nil
   end
   
   it "should provide rand, if it has a vector" do
-    Gaussian.should_receive(:rand).and_return(0.5)
+    Distributions::Gaussian.should_receive(:rand).and_return(0.5)
     @n.rand
   end
   
   it "should return nil for rand if it doesn't have a vector" do
-    @n = DummyNumericNode.new
+    @n = Distributions::DummyNumericNode.new
     @n.mean.should be_nil
   end
   
@@ -65,18 +65,60 @@ describe NumericMethods do
   end
   
   it "should pass the mean and standard deviation to the distribution when calculating an inverse_cdf" do
-    @n.inverse_cdf.should eql(Gaussian.inverse_cdf(:mean => @n.mean, :sd => @n.sd))
+    @n.inverse_cdf.should eql(Distributions::Gaussian.inverse_cdf(:mean => @n.mean, :sd => @n.sd))
   end
 
   it "should pass the mean and standard deviation to the distribution when calculating an lower_bound" do
-    @n.lower_bound.should eql(Gaussian.lower_bound(:mean => @n.mean, :sd => @n.sd))
+    @n.lower_bound.should eql(Distributions::Gaussian.lower_bound(:mean => @n.mean, :sd => @n.sd))
   end
 
   it "should pass the mean and standard deviation to the distribution when calculating an upper_bound" do
-    @n.upper_bound.should eql(Gaussian.upper_bound(:mean => @n.mean, :sd => @n.sd))
+    @n.upper_bound.should eql(Distributions::Gaussian.upper_bound(:mean => @n.mean, :sd => @n.sd))
   end
 
   it "should pass the mean and standard deviation to the distribution when calculating an interval_values" do
-    @n.interval_values.should eql(Gaussian.interval_values(:mean => @n.mean, :sd => @n.sd))
+    @n.interval_values.should eql(Distributions::Gaussian.interval_values(:mean => @n.mean, :sd => @n.sd))
   end
+  
+  it "should have a coefficient_of_variation" do
+    @n.coefficient_of_variation.should eql(@n.sd / @n.mean)
+  end
+  
+  it "should have a cov alias for coefficient_of_variation" do
+    @n.cov.should eql(@n.coefficient_of_variation)
+  end
+  
+  it "should have a summary hash" do
+    @n.should be_respond_to(:summary)
+    @n.summary.should be_a(Hash)
+  end
+  
+  it "should have a mean in the summary" do
+    @n.summary[:mean].should eql(@n.mean)
+  end
+  
+  it "should have a standard deviation in the summary" do
+    @n.summary[:standard_deviation].should eql(@n.standard_deviation)
+  end
+  
+  it "should have a min in the summary" do
+    @n.summary[:min].should eql(@n.vector.min)
+  end
+  
+  it "should have a max in the summary" do
+    @n.summary[:max].should eql(@n.vector.max)
+  end
+  
+  it "should have a lower_bound in the summary" do
+    @n.summary[:lower_bound].should eql(@n.lower_bound)
+  end
+  
+  it "should have an upper_bound in the summary" do
+    @n.summary[:upper_bound].should eql(@n.upper_bound)
+  end
+  
+  it "should have a coefficient_of_variation in the summary" do
+    @n.summary[:coefficient_of_variation].should eql(@n.sd / @n.mean)
+  end
+  
 end

data/spec/fathom/{plausible_range_spec.rb → node/plausible_range_spec.rb}

@@ -1,4 +1,4 @@
-require File.expand_path(File.dirname(__FILE__) + '/../spec_helper')
+require File.expand_path(File.dirname(__FILE__) + '/../../spec_helper')
 
 include Fathom
 

data/spec/fathom/node_spec.rb

@@ -151,5 +151,22 @@ describe Node do
     n2 = Node.new :name => 'n2', :parent => n1
     n1.n2.should eql(n2)
   end
+  
+  # it "should have introduced the Spira infrastructure by now" do
+  #   Object.should be_const_defined(:Spira)
+  # end
+  
+  # it "should have defined the spira repository" do
+  #   Spira.repositories[:default].should be_a(RDF::Repository)
+  # end
+
+  # This really changes the interface for everything.  So, either I re-purpose a node, or something...
+  # Next steps: read the Spira source and figure out what's easier to do: replace the functionality
+  # or change my interfaces. (btw, 9 issues, only.  
+  # Mostly just tied to creating the node accessors (n1.n2 type thing))
+  
+  # it "should have Spira included" do
+  #   Node.included_modules.should be_include(Spira::Resource)
+  # end
 
 end

data/spec/fathom_spec.rb

@@ -14,3 +14,43 @@ describe "Fathom" do
     Fathom.kb.should eql(Fathom.knowledge_base)
   end
 end
+
+
+=begin
+  OK, so I'm thinking about how this works....
+  A knowledge base is something I work from.  I could create a default knowledge base.  I won't be able to save it to anywhere permenant, but possibly I want to just be able to start working for a while, have things link up, then decide if I've created anything of value...
+  
+  Or, I could always use an explicit knowledge base.
+  
+  Frankly, the way I thought I would use it would be:
+  
+  home_budget = Fathom::KnowledgeBase.find_or_create(:home_budget)
+  home_budget.new(:plausible_range, :hard_min => 0, :min => 500, :max => 750, :name => "Referral Income", :type => :revenue)
+  revenue = home_budget.find(...)
+  mc_revenue = home_budget.new(:mc_node, :values => revenue, :type => :revenue, :name => "Combined Income", :description => "A collection of income from our investments, job, and miscellaneous projects.") do |s|
+    :combined_income => s.values.inject(0.0) do |sum, e| 
+      sum += e
+    end
+  end
+  home_budget.new(:svm_node, :values => mc_revenue.process, :type => :revenue)
+  
+  mc_daily_expenses = home_budget.new(:mc_node, 
+    :values => {:expenses => home_budget.find(...), :dates => Fathom::KnowledgeBase.find(:dates)},
+    :type => :expenses, 
+    :name => "Combined Expenses", 
+    :description => "All of the household expenses on a monthly basis"
+  ) do |s|
+    cycle ||= (1..365).cycle
+    day = s.values.dates.day_in_year(cycle.next)
+    ...
+  end
+  
+  create a mini belief network about whether the budget is appropriate here...
+  
+  
+  
+  For this demo, I picked up a project I had been working on.  I added a Monte Carlo analysis that 
+  could always be re-run, in case the inputs change.  I don't yet have an automated way to keep up with the
+  changes to the inputs on that, but I do on the belief nodes.  
+  
+=end