[Boost-commit] svn:boost r72911 - sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test

Subject: [Boost-commit] svn:boost r72911 - sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test
From: erwann.rogard_at_[hidden]
Date: 2011-07-05 07:23:17

Author: e_r
Date: 2011-07-05 07:23:16 EDT (Tue, 05 Jul 2011)
New Revision: 72911
URL: http://svn.boost.org/trac/boost/changeset/72911

Log:
cleaning up directory
Text files modified: 
   sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table1.cpp |   103 ++++++++++++++++---------               
   sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table2.cpp |   154 ++++++++++++++++++++++++--------------- 
   2 files changed, 161 insertions(+), 96 deletions(-)

Modified: sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table1.cpp
==============================================================================
--- sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table1.cpp	(original)
+++ sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table1.cpp	2011-07-05 07:23:16 EDT (Tue, 05 Jul 2011)
@@ -5,25 +5,30 @@
 //  Software License, Version 1.0. (See accompanying file                    //
 //  LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)         //
 ///////////////////////////////////////////////////////////////////////////////
+#include <boost/test/test_tools.hpp>
+#include <boost/test/unit_test.hpp>
+
 #include <string>
 #include <boost/mpl/int.hpp>
+#include <boost/mpl/vector/vector10.hpp>
+#include <boost/mpl/detail/wrapper.hpp>
+#include <boost/typeof/typeof.hpp>
 #include <boost/fusion/include/make_map.hpp>
 #include <boost/fusion/container/map/detail/sequence_to_map.hpp>
-#include <boost/mpl/vector/vector10.hpp>
 #include <boost/accumulators/framework/accumulator_set.hpp>
 #include <boost/accumulators/statistics/stats.hpp>
 #include <boost/assign/list_of.hpp>
-#include <boost/statistics/detail/non_parametric/contingency_table/include/pearson_chisq/independence.hpp>
 #include <boost/statistics/detail/non_parametric/contingency_table/include/factor.hpp>
 #include <boost/statistics/detail/non_parametric/contingency_table/include/cells.hpp>
+#include <boost/statistics/detail/non_parametric/contingency_table/include/pearson_chisq/independence.hpp>
 
-#include <libs/statistics/detail/non_parametric/test/contingency_table1.h>
+#include <boost/statistics/detail/non_parametric/contingency_table/pearson_chisq/independence/tag.hpp>
 
 void test_contingency_table1()
 {
-
+    namespace ac = boost::accumulators;
         namespace ct = boost::statistics::detail::contingency_table;
-    namespace pearson_chisq = ct::pearson_chisq;
+    namespace ps = ct::pearson_chi_square_statistic;
 
     typedef double val_;
     typedef boost::mpl::int_<0> x_; typedef int data_x_;
@@ -35,62 +40,84 @@
     >::type sample_;
     
     typedef boost::mpl::vector2<x_,y_> keys_;
-    typedef ct::pearson_chisq::tag::independence<keys_> hypothesis_;
-    typedef boost::accumulators::stats< hypothesis_> stats_;
-    typedef boost::accumulators::accumulator_set< sample_, stats_, long int > acc_;
+    typedef ps::tag::independence_between<keys_> hypothesis_;
+    typedef ac::stats< hypothesis_> stats_;
+    typedef ac::accumulator_set< sample_, stats_, long int > acc_;
         
     using namespace boost::assign;
-    acc_ acc(( ct::factor::_map_of_levels = boost::fusion::make_map<x_,y_>(
+    acc_ acc(( ct::_map_of_levels = boost::fusion::make_map<x_,y_>(
         list_of(-1)(1), list_of("a")("b")("c") ) ));
     
     {   
         // Check levels 
-     	BOOST_ASSERT( ct::cells::cells_count<keys_>( acc ) == 2 * 3 );
-     	BOOST_ASSERT( ct::factor::extract::levels<x_>( acc ).count( -1  ) == 1 );
-     	BOOST_ASSERT( ct::factor::extract::levels<x_>( acc ).count(  1  ) == 1 );
-     	BOOST_ASSERT( ct::factor::extract::levels<x_>( acc ).count(  2  ) == 0 );
-     	BOOST_ASSERT( ct::factor::extract::levels<y_>( acc ).count( "a" ) == 1 );
-     	BOOST_ASSERT( ct::factor::extract::levels<y_>( acc ).count( "b" ) == 1 );
-     	BOOST_ASSERT( ct::factor::extract::levels<y_>( acc ).count( "c" ) == 1 );
-     	BOOST_ASSERT( ct::factor::extract::levels<y_>( acc ).count( "d" ) == 0 );
+     	BOOST_CHECK( ct::cells_count<keys_>( acc ) == 2 * 3 );
+     	BOOST_CHECK( ct::extract::levels<x_>( acc ).count( -1  ) == 1 );
+     	BOOST_CHECK( ct::extract::levels<x_>( acc ).count(  1  ) == 1 );
+     	BOOST_CHECK( ct::extract::levels<x_>( acc ).count(  2  ) == 0 );
+     	BOOST_CHECK( ct::extract::levels<y_>( acc ).count( "a" ) == 1 );
+     	BOOST_CHECK( ct::extract::levels<y_>( acc ).count( "b" ) == 1 );
+     	BOOST_CHECK( ct::extract::levels<y_>( acc ).count( "c" ) == 1 );
+     	BOOST_CHECK( ct::extract::levels<y_>( acc ).count( "d" ) == 0 );
     }
     {                                                                        //.....|.....|.....|.....|.....|.....|.....|
                                                                              //     |non- |    x =    |       y =       |
                                                                              //total|empty|...........|.................|
         // Filling the cells                                                 //count|cells| -1  |  1  | "a" | "b" | "c" |
-        using namespace boost::accumulators;                                 //.....|count|.....|.....|.....|.....|.....|
+        using namespace ac;                                 //.....|count|.....|.....|.....|.....|.....|
         acc( boost::fusion::make_map<x_,y_,z_>( -1, "a", 1 ), weight = 1 );  //  1  |  1  |  1  |  0  |  1  |  0  |  0  |
         acc( boost::fusion::make_map<x_,y_,z_>(  1, "b", 1 ), weight = 2 );  //  3  |  2  |  1  |  2  |  1  |  2  |  0  |
     }
     {   // Check degrees of freedom
         // lost df = (r-1) + (c-1) = 1 + 2 = 3;
         //  df = rc -r -c + 1 = 3 * 2 - 3  -2 + 1 = 2 
-        BOOST_ASSERT( pearson_chisq::lost_df( acc, hypothesis_() ) == 3);
-        BOOST_ASSERT( pearson_chisq::df( acc, hypothesis_() ) == 2);
+        
+        typedef boost::mpl::detail::wrapper<hypothesis_> h0_;
+        BOOST_CHECK( ps::lost_degrees_of_freedom( h0_(), acc ) == 3 );
+        BOOST_CHECK( ps::degrees_of_freedom( h0_(),acc ) == 2 );
     }
     {
         // Check cells 
         typedef boost::mpl::vector1<x_> vec_x_;
         typedef boost::mpl::vector1<y_> vec_y_;   
         std::size_t n;
-        n = boost::accumulators::extract::weighted_count( acc );
-     	BOOST_ASSERT( n == 3 );
-        n = ct::cells::non_empty_cells_count<keys_>( acc );
-        BOOST_ASSERT( n == 2 );
-        n = ct::cells::extract::cells<vec_x_>( acc )[ boost::fusion::make_map<x_>( -1 ) ];
-     	BOOST_ASSERT( n == 1 );
-        n = ct::cells::extract::cells<vec_x_>( acc )[ boost::fusion::make_map<x_>( 1 ) ];
-     	BOOST_ASSERT( n == 2 );
-        n = ct::cells::extract::cells<vec_y_>( acc )[ boost::fusion::make_map<y_>( "a" ) ];
-     	BOOST_ASSERT( n == 1 );
-        n = ct::cells::extract::cells<vec_y_>( acc )[ boost::fusion::make_map<y_>( "b" ) ];
-     	BOOST_ASSERT( n == 2 );
-        n = ct::cells::extract::cells<vec_y_>( acc )[ boost::fusion::make_map<y_>( "c" ) ];
-     	BOOST_ASSERT( n == 0 );
-        n = ct::cells::extract::cells<keys_>( acc )[ boost::fusion::make_map<x_,y_,z_>( -1, "a", 1 ) ];
-        BOOST_ASSERT( n == 1 );
-        n = ct::cells::extract::cells<keys_>( acc )[ boost::fusion::make_map<x_,y_,z_>(  1, "b", 1 ) ];
-        BOOST_ASSERT( n == 2 );
+        n = ac::extract::weighted_count( acc );
+     	BOOST_CHECK( n == 3 );
+
+        n = ct::non_empty_cells_count<vec_x_>(acc);
+        BOOST_CHECK( n == 2 );
+        n = ct::non_empty_cells_count<vec_y_>(acc);
+        BOOST_CHECK( n == 2 );
+        n = ct::non_empty_cells_count<keys_>( acc );
+        BOOST_CHECK( n == 2 );
+
+        n = ct::cells_count<vec_x_>(acc);
+        BOOST_CHECK( n == 2 );
+        n = ct::cells_count<vec_y_>(acc);
+        BOOST_CHECK( n == 3 );
+        n = ct::cells_count<keys_>( acc );
+        BOOST_CHECK( n == 6 );
+        
+        n = ct::count_matching<vec_x_>( 
+            acc, boost::fusion::make_map<x_>( -1 ) );
+     	BOOST_CHECK( n == 1 );
+        n = ct::count_matching<vec_x_>( 
+            acc, boost::fusion::make_map<x_>( 1 ) );
+     	BOOST_CHECK( n == 2 );
+        n = ct::count_matching<vec_y_>( 
+            acc, boost::fusion::make_map<y_>( "a" ) );
+     	BOOST_CHECK( n == 1 );
+        n = ct::count_matching<vec_y_>( 
+            acc, boost::fusion::make_map<y_>( "b" ) );
+     	BOOST_CHECK( n == 2 );
+        n = ct::count_matching<vec_y_>( 
+            acc, boost::fusion::make_map<y_>( "c" ) );
+     	BOOST_CHECK( n == 0 );
+        n = ct::count_matching<keys_>( acc,
+            boost::fusion::make_map<x_,y_,z_>( -1, "a", 1 ) );
+        BOOST_CHECK( n == 1 );
+        n = ct::count_matching<keys_>( acc, 
+            boost::fusion::make_map<x_,y_,z_>(  1, "b", 1 ) );
+        BOOST_CHECK( n == 2 );
     }
 
 }

Modified: sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table2.cpp
==============================================================================
--- sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table2.cpp	(original)
+++ sandbox/statistics/non_parametric/libs/statistics/detail/non_parametric/test/contingency_table2.cpp	2011-07-05 07:23:16 EDT (Tue, 05 Jul 2011)
@@ -8,8 +8,13 @@
 #include <cmath>
 #include <string>
 
+#include <boost/test/unit_test.hpp>
+
 #include <boost/mpl/int.hpp>
 #include <boost/mpl/vector/vector10.hpp>
+#include <boost/mpl/detail/wrapper.hpp>
+
+#include <boost/typeof/typeof.hpp>
 
 #include <boost/assign/list_of.hpp>
 
@@ -23,8 +28,10 @@
 
 #include <boost/statistics/detail/non_parametric/contingency_table/include/factor.hpp>
 #include <boost/statistics/detail/non_parametric/contingency_table/include/pearson_chisq/independence.hpp>
+#include <boost/statistics/detail/non_parametric/contingency_table/include/pearson_chisq/common.hpp>
 
-#include <libs/statistics/detail/non_parametric/test/contingency_table2.h>
+// Analysis of a dataset, and check that the results agree with the R software.
+// Source : http://www.math.wustl.edu/~victor/classes/ma322/r-eg-20.txt
 
 /*
 # Read the count data for this problem
@@ -62,8 +69,7 @@
 {
 
         namespace ct = boost::statistics::detail::contingency_table;
-    namespace pearson_chisq = ct::pearson_chisq;
-    using namespace boost::accumulators;
+    namespace ps = ct::pearson_chi_square_statistic;
 
     typedef double val_;
     typedef boost::mpl::int_<0> r_; typedef std::string data_r_;
@@ -79,10 +85,10 @@
     typedef boost::mpl::vector2<r_,t_> sum_over_cols_;
     typedef boost::mpl::vector2<c_,t_> sum_over_rows_;
 
-    typedef ct::pearson_chisq::tag::independence<all_three_factors_> indep_r_c_t_;
-    typedef ct::pearson_chisq::tag::independence<sum_over_tiers_> indep_r_c_;
-    typedef ct::pearson_chisq::tag::independence<sum_over_cols_> indep_r_t_;
-    typedef ct::pearson_chisq::tag::independence<sum_over_rows_> indep_c_t_;
+    typedef ps::tag::independence_between<all_three_factors_> indep_r_c_t_;
+    typedef ps::tag::independence_between<sum_over_tiers_> indep_r_c_;
+    typedef ps::tag::independence_between<sum_over_cols_> indep_r_t_;
+    typedef ps::tag::independence_between<sum_over_rows_> indep_c_t_;
 
     typedef boost::accumulators::stats<
         indep_r_c_t_,
@@ -93,8 +99,11 @@
     typedef boost::accumulators::accumulator_set< sample_, stats_, long int > acc_;
         
     using namespace boost::assign;
-    acc_ acc(( ct::factor::_map_of_levels = boost::fusion::make_map<r_,c_,t_>(
-        list_of("r1")("r2")("r3")("r4"), list_of("c1")("c2")("c3"), list_of("t1")("t2") ) ));
+    acc_ acc(( ct::_map_of_levels = boost::fusion::make_map<r_,c_,t_>(
+        list_of("r1")("r2")("r3")("r4"), 
+        list_of("c1")("c2")("c3"), 
+        list_of("t1")("t2") ) 
+    ));
 
     typedef boost::fusion::result_of::make_map<
         r_,c_,t_,data_r_,data_c_,data_t_>::type result_of_make_map_;
@@ -102,35 +111,38 @@
     
     fp_ make_sample = boost::fusion::make_map<r_,c_,t_>;
 
-    acc( make_sample( "r1", "c1", "t1" ), weight = 12 );
-    acc( make_sample( "r1", "c1", "t2" ), weight =  4 );
-    acc( make_sample( "r1", "c2", "t1" ), weight = 34 );
-    acc( make_sample( "r1", "c2", "t2" ), weight = 47 );
-    acc( make_sample( "r1", "c3", "t1" ), weight = 23 );
-    acc( make_sample( "r1", "c3", "t2" ), weight = 11 );
-    acc( make_sample( "r2", "c1", "t1" ), weight = 35 );
-    acc( make_sample( "r2", "c1", "t2" ), weight = 34 );
-    acc( make_sample( "r2", "c2", "t1" ), weight = 31 );
-    acc( make_sample( "r2", "c2", "t2" ), weight = 10 );
-    acc( make_sample( "r2", "c3", "t1" ), weight = 11 );
-    acc( make_sample( "r2", "c3", "t2" ), weight = 18 );
-    acc( make_sample( "r3", "c1", "t1" ), weight = 12 );
-    acc( make_sample( "r3", "c1", "t2" ), weight = 18 );
-    acc( make_sample( "r3", "c2", "t1" ), weight = 32 );
-    acc( make_sample( "r3", "c2", "t2" ), weight = 13 );
-    acc( make_sample( "r3", "c3", "t1" ), weight =  9 );
-    acc( make_sample( "r3", "c3", "t2" ), weight = 19 );
-    acc( make_sample( "r4", "c1", "t1" ), weight = 12 );
-    acc( make_sample( "r4", "c1", "t2" ), weight =  9 );
-    acc( make_sample( "r4", "c2", "t1" ), weight = 12 );
-    acc( make_sample( "r4", "c2", "t2" ), weight = 33 );
-    acc( make_sample( "r4", "c3", "t1" ), weight = 14 );
-    acc( make_sample( "r4", "c3", "t2" ), weight = 25 );
+    { 
+        using namespace boost::accumulators;
 
-    using namespace std;
+        acc( make_sample( "r1", "c1", "t1" ), weight = 12 );
+        acc( make_sample( "r1", "c1", "t2" ), weight =  4 );
+        acc( make_sample( "r1", "c2", "t1" ), weight = 34 );
+        acc( make_sample( "r1", "c2", "t2" ), weight = 47 );
+        acc( make_sample( "r1", "c3", "t1" ), weight = 23 );
+        acc( make_sample( "r1", "c3", "t2" ), weight = 11 );
+        acc( make_sample( "r2", "c1", "t1" ), weight = 35 );
+        acc( make_sample( "r2", "c1", "t2" ), weight = 34 );
+        acc( make_sample( "r2", "c2", "t1" ), weight = 31 );
+        acc( make_sample( "r2", "c2", "t2" ), weight = 10 );
+        acc( make_sample( "r2", "c3", "t1" ), weight = 11 );
+        acc( make_sample( "r2", "c3", "t2" ), weight = 18 );
+        acc( make_sample( "r3", "c1", "t1" ), weight = 12 );
+        acc( make_sample( "r3", "c1", "t2" ), weight = 18 );
+        acc( make_sample( "r3", "c2", "t1" ), weight = 32 );
+        acc( make_sample( "r3", "c2", "t2" ), weight = 13 );
+        acc( make_sample( "r3", "c3", "t1" ), weight =  9 );
+        acc( make_sample( "r3", "c3", "t2" ), weight = 19 );
+        acc( make_sample( "r4", "c1", "t1" ), weight = 12 );
+        acc( make_sample( "r4", "c1", "t2" ), weight =  9 );
+        acc( make_sample( "r4", "c2", "t1" ), weight = 12 );
+        acc( make_sample( "r4", "c2", "t2" ), weight = 33 );
+        acc( make_sample( "r4", "c3", "t1" ), weight = 14 );
+        acc( make_sample( "r4", "c3", "t2" ), weight = 25 );
+    }
 
-    val_ stat;
-    long df;
+    using namespace std;
+    val_ stat, p_value;
+    long df;    
 
 /*
 #### Output and interpretation (at significance level alpha=0.05):
@@ -147,13 +159,18 @@
 #
 */
 
-    // #include <boost/test/unit_test.hpp>
-    // #include <boost/test/floating_point_comparison.hpp>
-
-    df = pearson_chisq::df(acc,indep_r_c_t_());
-    stat = pearson_chisq::statistic<val_>( acc, indep_r_c_t_() );
-    BOOST_ASSERT( df == 17 );
-    BOOST_ASSERT( fabs(stat - 102.17) < 0.01 );
+   {
+        typedef boost::mpl::detail::wrapper<indep_r_c_t_> h0_;
+        df = ps::degrees_of_freedom( h0_(), acc);
+        stat = ps::value( h0_(), acc ); 
+        p_value = cdf( complement( 
+            ps::asy_distribution( h0_(), acc ), 
+            stat
+        ) );
+        BOOST_CHECK( df == 17 );
+        BOOST_CHECK( fabs( stat - 102.17 ) < 0.01 );
+        BOOST_CHECK( fabs( p_value - 3.514e-14 ) < 0.001e-14 );
+    }
 
 /*
    > # 2-way Chi squared test of partial independence: R versus CT
@@ -168,11 +185,18 @@
 #
 */
 
-    df = pearson_chisq::df(acc,indep_c_t_());
-    stat = pearson_chisq::statistic<val_>( acc, indep_c_t_() );
-    BOOST_ASSERT( df == 2 );
-    BOOST_ASSERT( fabs( stat == 2.3704 )< 0.001 );
-
+   {
+        typedef boost::mpl::detail::wrapper<indep_c_t_> h0_;
+        df = ps::degrees_of_freedom( h0_(), acc);
+        stat = ps::value( h0_(), acc ); 
+        p_value = cdf( complement( 
+            ps::asy_distribution( h0_(), acc ), 
+            stat
+        ) );
+        BOOST_CHECK( df == 2 );
+        BOOST_CHECK( fabs( stat == 2.3704 )< 0.001 );
+        BOOST_CHECK( fabs( p_value - 0.3057 ) < 0.0001 );
+    }
 /*
    >
    > # 2-way Chi squared test of partial independence: C versus RT
@@ -187,12 +211,18 @@
 #            HA: r,t are NOT mutually independent
 
 */
-
-    df = pearson_chisq::df(acc,indep_r_t_());
-    stat = pearson_chisq::statistic<val_>( acc, indep_r_t_() );
-    BOOST_ASSERT( df == 3 );
-    BOOST_ASSERT( fabs( stat - 10.057 ) < 0.001 );
-
+    {
+        typedef boost::mpl::detail::wrapper<indep_r_t_> h0_;
+        df = ps::degrees_of_freedom( h0_(), acc);
+        stat = ps::value( h0_(), acc ); 
+        p_value = cdf( complement( 
+            ps::asy_distribution( h0_(), acc ), 
+            stat
+        ) );
+        BOOST_CHECK( df == 3 );
+        BOOST_CHECK( fabs( stat - 10.057 ) < 0.001 );
+        BOOST_CHECK( fabs( p_value - 0.01809 ) < 0.00001 );
+    }
 /*
    > # 2-way Chi squared test of partial independence: T versus RC
    > summary(xtabs(count~r+c))
@@ -208,10 +238,18 @@
 #
 */
 
-    df = pearson_chisq::df(acc,indep_r_c_());
-    stat = pearson_chisq::statistic<val_>( acc, indep_r_c_() );
-    BOOST_ASSERT( df == 6 );
-    BOOST_ASSERT( fabs( stat - 58.67 ) < 0.01 );
+   {
+        typedef boost::mpl::detail::wrapper<indep_r_c_> h0_;
+        df = ps::degrees_of_freedom( h0_(), acc);
+        stat = ps::value( h0_(), acc ); 
+        p_value = cdf( complement( 
+            ps::asy_distribution( h0_(), acc ), 
+            stat
+        ) );
+        BOOST_CHECK( df == 6 );
+        BOOST_CHECK( fabs( stat - 58.67 ) < 0.01 );
+        BOOST_CHECK( fabs( p_value - 8.363e-11 ) < 0.001e-11 );
+    }
 
 }