statsample-sem 0.1.0

data/spec/spec.opts

@@ -0,0 +1,3 @@
+--color
+-f s 
+-b

data/spec/spec_helper.rb

@@ -0,0 +1,10 @@
+# encoding: utf-8
+
+puts "Statsample-SEM specs: Running on Ruby Version: #{RUBY_VERSION}"
+
+require "rubygems"
+require 'spec'
+require 'spec/autorun'
+require 'statsample'
+$LOAD_PATH.unshift File.join(File.dirname(__FILE__), '..', 'lib') 
+require 'statsample/sem'

data/spec/statsample_sem_model_spec.rb

@@ -0,0 +1,158 @@
+$:.unshift(File.dirname(__FILE__)+"/.")
+require 'spec_helper'
+
+describe Statsample::SEM::Model do
+  before(:all) do
+    begin 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.ds"
+      @ds=Statsample.load(@data_path)
+    rescue 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.csv"
+      @ds=Statsample::CSV.read(@data_path)
+    end
+  end
+
+  before(:each) do
+    @model=Statsample::SEM::Model.new
+  end
+  it "method manifests= should set manifests" do
+    @model.manifests=@ds.fields
+    @model.manifests.should==@ds.fields
+  end
+  it "method manifests with argument should set manifests" do
+    @model.manifests @ds.fields
+    @model.manifests.should==@ds.fields
+  end
+  it "method latents= should set latents" do
+    @model.latents=["G"]
+    @model.latents.should==["G"]
+  end
+  it "method latents with argument should set latents" do
+    @model.latents ["G"]
+    @model.latents.should==["G"]
+  end
+  it "should automaticly set manifest and latents with data and path" do
+      @model.data_from_matrix(Statsample::Bivariate.covariance_matrix(@ds), :cases=>@ds.dup_only_valid.cases)
+      @model.path :from=>"G1", :to=>['x1','x2']
+      @model.path :from=>"G2", :to=>['x3','x4']
+      @model.latents.should==['G1','G2']
+      @model.manifests.should==['x1','x2','x3','x4']
+  end
+  
+  
+  
+  it "method path with :from should create a variance path" do
+    @model.path :from=>"x1"
+    @model.paths.should=={
+      ['x1','x1']=>{:from=>'x1',:to=>'x1',:arrow=>2,:label=>'var x1', :free=>true, :value=>nil}
+    }
+  end
+  it "method path with :from and :two should create a regression path" do
+    @model.path :from=>"G", :to=>'x1'
+    @model.paths.should=={
+      ['G','x1']=>{:from=>'G',:to=>'x1',:arrow=>1,:label=>'G to x1', :free=>true, :value=>nil}
+    }
+  end
+  it "method path with :from, :two and arrows should set label automaticly" do
+    @model.path :from=>"G", :to=>'x1', :arrows=>1
+    @model.path :from=>"G", :to=>'x2', :arrows=>2
+
+    @model.paths.should=={
+      ['G','x1']=>{:from=>'G',:to=>'x1',:arrow=>1,:label=>'G to x1', :free=>true, :value=>nil},
+      ['G','x2']=>{:from=>'G',:to=>'x2',:arrow=>2,:label=>'G cov x2', :free=>true, :value=>nil}
+    }
+  end 
+  it "method path should label correctly" do
+    @model.path :from=>"G", :to=>['x1','x2','x3'], :labels=>['to x1','to x2']
+    @model.paths.should=={
+      ['G','x1']=>{:from=>'G',:to=>'x1',:arrow=>1,:label=>'to x1', :free=>true, :value=>nil},
+      ['G','x2']=>{:from=>'G',:to=>'x2',:arrow=>1,:label=>'to x2', :free=>true, :value=>nil},
+      ['G','x3']=>{:from=>'G',:to=>'x3',:arrow=>1,:label=>'G to x3', :free=>true, :value=>nil}
+    }
+  end 
+  
+  
+  
+  it "method path with free array set values in proper order" do
+    @model.path :from=>"G", :to=>['x1','x2','x3'], :free=>[false,true,false], :values=>[1,2,3]
+    @model.paths.should=={
+      ['G','x1']=>{:from=>'G',:to=>'x1',:arrow=>1,:label=>'G to x1', :free=>false, :value=>1},
+      ['G','x2']=>{:from=>'G',:to=>'x2',:arrow=>1,:label=>'G to x2', :free=>true, :value=>nil},
+      ['G','x3']=>{:from=>'G',:to=>'x3',:arrow=>1,:label=>'G to x3', :free=>false, :value=>3}
+    }
+  end
+  it "method path with free array set scalar value in proper order" do
+    @model.path :from=>"G", :to=>['x1','x2','x3'], :free=>[false,true,false], :values=>1.0
+    @model.paths.should=={
+      ['G','x1']=>{:from=>'G',:to=>'x1',:arrow=>1,:label=>'G to x1', :free=>false, :value=>1.0},
+      ['G','x2']=>{:from=>'G',:to=>'x2',:arrow=>1,:label=>'G to x2', :free=>true, :value=>nil},
+      ['G','x3']=>{:from=>'G',:to=>'x3',:arrow=>1,:label=>'G to x3', :free=>false, :value=>1.0}
+    }
+  end
+  
+  
+  describe "with multiple in and outs" do
+    it "should set correctly 1 in, multiple outs" do
+      @model.path :from=>"G", :to=>['x1','x2']
+      @model.paths.should=={
+        ['G','x1']=>{:from=>'G',:to=>'x1',:arrow=>1,:label=>'G to x1', :free=>true, :value=>nil},
+        ['G','x2']=>{:from=>'G',:to=>'x2',:arrow=>1,:label=>'G to x2', :free=>true, :value=>nil}
+      }
+    end
+    it "should set correctly multiple ins, 1 out" do
+      @model.path :from=>["x1","x2"], :to=>['x3']
+      @model.paths.should=={
+        ['x1','x3']=>{:from=>'x1',:to=>'x3',:arrow=>1,:label=>'x1 to x3', :free=>true, :value=>nil},
+        ['x2','x3']=>{:from=>'x2',:to=>'x3',:arrow=>1,:label=>'x2 to x3', :free=>true, :value=>nil}
+      }
+    end
+    it "should set correctly multiple ins, multiples outs, with :all=>false" do
+      @model.path :from=>["x1","x2"], :to=>['x3','x4']
+      @model.paths.should=={
+        ['x1','x3']=>{:from=>'x1',:to=>'x3',:arrow=>1,:label=>'x1 to x3', :free=>true, :value=>nil},
+        ['x2','x4']=>{:from=>'x2',:to=>'x4',:arrow=>1,:label=>'x2 to x4', :free=>true, :value=>nil}
+      }
+    end
+    it "should set correctly multiple ins, multiples outs, with :all=>true" do
+      @model.path :from=>["x1","x2"], :to=>['x3','x4'], :all=>true
+      @model.paths.should=={
+        ['x1','x3']=>{:from=>'x1',:to=>'x3',:arrow=>1,:label=>'x1 to x3', :free=>true, :value=>nil},
+        ['x1','x4']=>{:from=>'x1',:to=>'x4',:arrow=>1,:label=>'x1 to x4', :free=>true, :value=>nil},
+        ['x2','x3']=>{:from=>'x2',:to=>'x3',:arrow=>1,:label=>'x2 to x3', :free=>true, :value=>nil},
+
+        ['x2','x4']=>{:from=>'x2',:to=>'x4',:arrow=>1,:label=>'x2 to x4', :free=>true, :value=>nil}
+      }
+    end
+  end
+  
+  it "should accept a dataset" do
+    @model.data_from_dataset(@ds)
+    @model.data_type.should==:raw
+    @model.ds.should==@ds
+    @model.variables.should==@ds.fields
+  end
+  it "should accept a covariance matrix" do
+    @model.data_from_matrix(Statsample::Bivariate.covariance_matrix(@ds), :cases=>@ds.dup_only_valid.cases)
+    @model.data_type.should==:covariance
+    @model.cases.should==@ds.cases
+    @model.variables.should==@ds.fields
+  end
+  it "should accept a correlation matrix" do
+    @model.data_from_matrix(Statsample::Bivariate.correlation_matrix(@ds), :cases=>@ds.dup_only_valid.cases)
+    @model.data_type.should==:correlation
+    @model.cases.should==@ds.cases
+    @model.variables.should==@ds.fields
+  end
+  it "should duplicate correctly using 'dup'" do
+    @model.data_from_dataset(@ds)
+    @model.path :from=>"G", :to=>['x1','x2']
+    model2=@model.dup
+    model2.paths.should==@model.paths
+    model2.manifests.should==@model.manifests
+    model2.latents.should==@model.latents
+    model2.cases.should==@model.cases
+    
+    
+  end
+  
+end

data/spec/statsample_sem_openmxengine_spec.rb

@@ -0,0 +1,142 @@
+$:.unshift(File.dirname(__FILE__)+"/.")
+require 'spec_helper'
+
+describe Statsample::SEM::OpenMxEngine do
+  before(:all) do
+    begin 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.ds"
+      @ds=Statsample.load(@data_path)
+    rescue 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.csv"
+      @ds=Statsample::CSV.read(@data_path)
+    end
+    @cov_matrix=Statsample::Bivariate.covariance_matrix(@ds)
+    @cases=@ds.cases
+  end
+  describe "using matrix based data" do 
+    before(:all) do
+      @model=Statsample::SEM::Model.new do |m|
+        m.manifests @ds.fields
+        m.latents %w{G}
+        m.path :from=>m.latents, :to=>m.manifests
+        m.path :from=>m.manifests
+        m.path :from=>m.latents, :free=>false, :values=>1.0
+        m.data_from_matrix(@cov_matrix,:cases=>@cases)
+      end
+      @engine=Statsample::SEM::OpenMxEngine.new(@model)
+    end
+    it "should generate a valid r query" do
+      @engine.r_query.size.should>=0
+    end
+    it "should compute and return well formed response" do
+      lambda{@engine.compute}.should_not raise_error
+      pp @engine.r_summary
+      @engine.r_summary.should be_instance_of (Array)
+    end
+    it "should return a valid graphviz definition for model" do
+      @engine.graphviz.size.should>=0
+    end
+    it "method chi_square return X^2 of model" do 
+      @engine.chi_square.should be_close(7.384,0.001)
+    end
+    it "method df return degrees of freedom of model" do 
+      @engine.df.should==5
+    end
+    it "method chi_square_null return X^2 of null model" do
+      @engine.chi_square_null.should be_close(3725.0596,0.001)
+    end
+    it "method df_null return degrees of freedom for null model" do
+      @engine.df_null.should==10
+    end
+    it "method goodness_of_fit return GFI" do
+      pending('Not Implemented')
+      @engine.goodness_of_fit.should be_close(0.99426, 0.0001)
+    end
+    it "method adjusted_goodness_of_fit return AGFI" do
+      pending('Not Implemented')
+      @engine.adjusted_goodness_of_fit.should be_close(0.98277, 0.0001)
+    end
+    it "method rmsea and derivatives returns... RMSEA!" do
+      @engine.rmsea.should be_close(0.030911, 0.0001)
+    end
+    it "method rmsea_alpha and rmsea_confidence_interval returns alpa and ci for RMSEA" do 
+      pending('Not Implemented other methods')
+      @engine.rmsea_alpha.should==0.90
+      @engine.rmsea_confidence_interval[0].should be_nil
+      @engine.rmsea_confidence_interval[1].should be_close(0.074569, 0.0001)
+    end
+    it "method nfi returns NFI" do
+      @engine.nfi.should be_close(0.99802, 0.0001)
+    end
+    it "method nnfi returns NNFI" do
+     
+      @engine.nnfi.should be_close(0.99872, 0.0001)
+    end
+    it "method cfi returns CFI" do
+      @engine.cfi.should be_close(0.99936, 0.0001)
+    end
+    it "method srmr returns SRMR" do
+      pending('Not Implemented')
+      @engine.srmr.should be_close(0.0032191, 0.0001)
+    end
+    it "method bic returns BIC" do
+      pending("Not well implemented")
+      @engine.bic.should be_close(-23.689, 0.001)
+    end
+    it "method iterations returns the number of iterations" do
+      pending('Not Implemented')
+      @engine.iterations.should==30
+    end
+    
+    # G to x1 0.397152 0.0155826 25.4869 0        x1 <--- G 
+    # G to x2 0.503662 0.0182767 27.5576 0        x2 <--- G 
+    # G to x3 0.577242 0.0204998 28.1584 0        x3 <--- G 
+    # G to x4 0.702774 0.0240755 29.1905 0        x4 <--- G 
+    # G to x5 0.796251 0.0267431 29.7741 0        x5 <--- G 
+    # var x1  0.040814 0.0028254 14.4455 0        x1 <--> x1
+    # var x2  0.038020 0.0028185 13.4893 0        x2 <--> x2
+    # var x3  0.040827 0.0031653 12.8986 0        x3 <--> x3
+    # var x4  0.039387 0.0034217 11.5110 0        x4 <--> x4
+    # var x5  0.036287 0.0036916  9.8296 0        x5 <--> x5    
+    it "coefficients returns a hash of Parameter estimates" do
+      coeffs=@engine.coefficients
+      coeffs[['G','x1']][:estimate].should be_close(0.397152, 0.00001)
+      coeffs[['G','x1']][:se].should be_close(0.0155826, 0.0001)
+      coeffs[['G','x1']][:label].should=='G to x1'
+     # pending('Not Implemented z')
+     # coeffs[['G','x1']][:z].should be_close(25.4869, 0.0001)
+    end
+    
+    
+  end
+  if(false)
+  describe "using raw data" do 
+    before(:each) do
+      @model=Statsample::SEM::Model.new do |m|
+        m.manifests @ds.fields
+        m.latents %w{G}
+        m.path :from=>m.latents, :to=>m.manifests
+        m.path :from=>m.manifests
+        m.path :from=>m.latents, :free=>false, :values=>1.0
+        m.path :from=>"one", :to=>m.manifests
+        m.data_from_dataset(@ds)
+      end
+      @engine=Statsample::SEM::OpenMxEngine.new(@model)
+    end
+    it "should generate a valid r query" do
+      @engine.r_query.size.should>=0
+    end
+    it "should compute and return well formed response" do
+      lambda{@engine.compute}.should_not raise_error
+      @engine.r_summary.should be_instance_of (Array)
+    end
+    it "should return a valid graphviz definition for model" do
+      @engine.graphviz.should match "digraph"
+    end
+    
+    
+  end
+  end
+  
+end
+

data/spec/statsample_sem_semjfoxengine_spec.rb

@@ -0,0 +1,116 @@
+$:.unshift(File.dirname(__FILE__)+"/.")
+require 'spec_helper'
+
+describe Statsample::SEM::SemJFoxEngine do
+  before(:all) do
+    begin 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.ds"
+      @ds=Statsample.load(@data_path)
+    rescue 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.csv"
+      @ds=Statsample::CSV.read(@data_path)
+    end
+    @cov_matrix=Statsample::Bivariate.covariance_matrix(@ds)
+    @cases=@ds.cases
+  end
+  describe "using matrix based data" do 
+    before(:all) do
+      @model=Statsample::SEM::Model.new do |m|
+        m.manifests @ds.fields
+        m.latents %w{G}
+        m.path :from=>m.latents, :to=>m.manifests
+        m.path :from=>m.manifests
+        m.path :from=>m.latents, :free=>false, :values=>1.0
+        m.data_from_matrix(@cov_matrix,:cases=>@cases)
+      end
+      @engine=Statsample::SEM::SemJFoxEngine.new(@model)
+    end
+    it "should generate a valid r query" do
+      @engine.r_query.size.should>=0
+    end
+    it "should compute and return well formed response" do
+      lambda{@engine.compute}.should_not raise_error
+      @engine.r_summary.should be_instance_of (Array)
+    end
+    it "should return a valid graphviz definition for model" do
+      @engine.graphviz.should match "digraph"
+    end
+    
+# Model Chisquare =  7.384   Df =  5 Pr(>Chisq) = 0.19361
+# Chisquare (null model) =  3725.1   Df =  10
+# Goodness-of-fit index =  0.99426
+# Adjusted goodness-of-fit index =  0.98277
+# RMSEA index =  0.030911   90% CI: (NA, 0.074569)
+# Bentler-Bonnett NFI =  0.99802
+# Tucker-Lewis NNFI =  0.99872
+# Bentler CFI =  0.99936
+# SRMR =  0.0032191
+# BIC =  -23.689 
+    
+    it "method chi_square return X^2 of model" do 
+      @engine.chi_square.should be_close(7.384,0.001)
+    end
+    it "method df return degrees of freedom of model" do 
+      @engine.df.should==5
+    end
+    it "method chi_square_null return X^2 of null model" do 
+      @engine.chi_square_null.should be_close(3725.0596,0.001)
+    end
+    it "method df_null return degrees of freedom for null model" do
+      @engine.df_null.should==10
+    end
+    it "method goodness_of_fit return GFI" do
+      @engine.goodness_of_fit.should be_close(0.99426, 0.0001)
+    end
+    it "method adjusted_goodness_of_fit return AGFI" do
+      @engine.adjusted_goodness_of_fit.should be_close(0.98277, 0.0001)
+    end
+    it "method rmsa and derivatives returns... RMSEA!" do
+      @engine.rmsea.should be_close(0.030911, 0.0001)
+      @engine.rmsea_alpha.should==0.90
+      @engine.rmsea_confidence_interval[0].should be_nil
+      @engine.rmsea_confidence_interval[1].should be_close(0.074569, 0.0001)
+    end
+    it "method nfi returns NFI" do
+      @engine.nfi.should be_close(0.99802, 0.0001)
+    end
+    it "method nnfi returns NNFI" do
+      @engine.nnfi.should be_close(0.99872, 0.0001)
+    end
+    it "method cfi returns CFI" do
+      @engine.cfi.should be_close(0.99936, 0.0001)
+    end
+    it "method srmr returns SRMR" do
+      @engine.srmr.should be_close(0.0032191, 0.0001)
+    end
+    it "method bic returns BIC" do
+      @engine.bic.should be_close(-23.689, 0.001)
+    end
+    it "method iterations returns the number of iterations" do
+      @engine.iterations.should==30
+    end
+    
+    # G to x1 0.397152 0.0155826 25.4869 0        x1 <--- G 
+    # G to x2 0.503662 0.0182767 27.5576 0        x2 <--- G 
+    # G to x3 0.577242 0.0204998 28.1584 0        x3 <--- G 
+    # G to x4 0.702774 0.0240755 29.1905 0        x4 <--- G 
+    # G to x5 0.796251 0.0267431 29.7741 0        x5 <--- G 
+    # var x1  0.040814 0.0028254 14.4455 0        x1 <--> x1
+    # var x2  0.038020 0.0028185 13.4893 0        x2 <--> x2
+    # var x3  0.040827 0.0031653 12.8986 0        x3 <--> x3
+    # var x4  0.039387 0.0034217 11.5110 0        x4 <--> x4
+    # var x5  0.036287 0.0036916  9.8296 0        x5 <--> x5    
+    it "coefficients returns a hash of Parameter estimates" do
+      coeffs=@engine.coefficients
+      coeffs[['G','x1']][:estimate].should be_close(0.397152, 0.00001)
+      coeffs[['G','x1']][:se].should be_close(0.0155826, 0.00001)
+      coeffs[['G','x1']][:z].should be_close(25.4869, 0.0001)
+      coeffs[['G','x1']][:label].should=='G to x1'
+    end
+    
+    
+  end
+  
+  
+end
+

data/spec/statsample_sem_spec.rb

@@ -0,0 +1,36 @@
+$:.unshift(File.dirname(__FILE__)+"/.")
+require 'spec_helper'
+
+describe Statsample::SEM do
+  before(:all) do
+    begin 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.ds"
+      @ds=Statsample.load(@data_path)
+    rescue 
+      @data_path=File.dirname(__FILE__)+"/fixtures/demo_open_mx.csv"
+      @ds=Statsample::CSV.read(@data_path)
+    end
+    @cov_matrix=Statsample::Bivariate.covariance_matrix(@ds)
+    @cases=@ds.cases
+    @sem=Statsample::SEM.new do |m|
+      m.manifests @ds.fields
+      m.latents %w{G}
+      m.path :from=>m.latents, :to=>m.manifests
+      m.path :from=>m.manifests
+      m.path :from=>m.latents, :free=>false, :values=>1.0
+      m.data_from_matrix(@cov_matrix,:cases=>@cases)
+    end
+  end
+  it "should calculate SEM using openmx" do
+    @sem.engine=:openmx
+    @sem.name="Using OpenMx"
+    @sem.compute.should be_true
+  end
+  it "should calculate SEM using sem" do
+    @sem.engine=:sem
+    @sem.name="Using SEM"
+    @sem.compute.should be_true
+  end
+  
+end
+