[Boost-commit] svn:boost r45739 - sandbox/cluster/libs/algorithm/cluster/test

From: yuanming_at_[hidden]
Date: 2008-05-25 02:52:58

Author: lakeeast
Date: 2008-05-25 02:52:58 EDT (Sun, 25 May 2008)
New Revision: 45739
URL: http://svn.boost.org/trac/boost/changeset/45739

Log:
Add unit test for k_means algorithm.
Added:
   sandbox/cluster/libs/algorithm/cluster/test/
   sandbox/cluster/libs/algorithm/cluster/test/Jamfile.v2   (contents, props changed)
   sandbox/cluster/libs/algorithm/cluster/test/k_means_test.cpp   (contents, props changed)

Added: sandbox/cluster/libs/algorithm/cluster/test/Jamfile.v2
==============================================================================
--- (empty file)
+++ sandbox/cluster/libs/algorithm/cluster/test/Jamfile.v2	2008-05-25 02:52:58 EDT (Sun, 25 May 2008)
@@ -0,0 +1,22 @@
+# Boost.Algorithm.Cluster Library
+
+# Copyright Yuanming Chen 2008. Use, modification and
+# distribution is subject to the Boost Software License, Version
+# 1.0. (See accompanying file LICENSE_1_0.txt or copy at
+# http://www.boost.org/LICENSE_1_0.txt)
+
+# For more information, see http://www.boost.org
+
+# bring in rules for testing
+import testing ;
+
+project
+    : source-location .
+    : requirements
+    ;
+{
+  test-suite cluster_test
+    :
+  [ run k_means_test.cpp ]
+ ;
+}

Added: sandbox/cluster/libs/algorithm/cluster/test/k_means_test.cpp
==============================================================================
--- (empty file)
+++ sandbox/cluster/libs/algorithm/cluster/test/k_means_test.cpp	2008-05-25 02:52:58 EDT (Sun, 25 May 2008)
@@ -0,0 +1,97 @@
+// Boost.Algorithm.Cluster Library
+// Copyright Yuanming Chen 2008. Use, modification and
+// distribution is subject to the Boost Software License, Version
+// 1.0. (See accompanying file LICENSE_1_0.txt or copy at
+// http://www.boost.org/LICENSE_1_0.txt)
+
+// For more information, see http://www.boost.org
+
+#include <boost/test/unit_test.hpp>
+#define BOOST_TEST_MODULE ClusterLib
+
+#include <boost/algorithm/cluster/k_means.hpp>
+#include <boost/array.hpp>
+#include <iostream>
+#include <fstream>
+#include <string> 
+using namespace boost::algorithm::cluster;
+
+BOOST_AUTO_TEST_CASE( basic_k_means_test )
+{
+   //The example can be found at:
+    //http://people.revoledu.com/kardi/tutorial/kMean/NumericalExample.htm
+
+    //Attribute 1: weight index.
+    //Attribute 2: pH value
+    //"POINT" represents a data point in the multi-dimensional data base
+    typedef boost::array<double, 2> POINT;
+    ////////////////////////////////////////////////////////////////////////
+    //Four drugs. We need to get two clusters
+    std::vector<POINT> vecPoints;
+
+    POINT medicineA;
+    medicineA[0] = 1;
+    medicineA[1] = 1;
+    vecPoints.push_back(medicineA);
+
+    POINT medicineB;
+    medicineB[0] = 2;
+    medicineB[1] = 1;
+    vecPoints.push_back(medicineB);
+
+    POINT medicineC;
+    medicineC[0] = 4;
+    medicineC[1] = 3;
+    vecPoints.push_back(medicineC);
+
+    POINT medicineD;
+    medicineD[0] = 5;
+    medicineD[1] = 4;
+    vecPoints.push_back(medicineD);
+
+    const int k = 2; //2 groups of medicine
+
+    typedef KMeansCluster<POINT> ClusterType;
+    typedef KMeansClustering<ClusterType> ClusteringType;
+    size_t knDimension = POINT::size();
+
+    ClusteringType result = k_means(vecPoints.begin(), vecPoints.end(), k, 0.00001);
+    int nCentroidId = 0;
+    for(ClusteringType::type::const_iterator iter = result.clusters.begin();
+        iter != result.clusters.end();
+        iter++, nCentroidId++)
+    {
+        const ClusterType& cluster = *iter;
+        std::cout << "Cluster " << nCentroidId << std::endl;
+        std::cout << "Centroid point (";
+
+        for(unsigned int nAttribute = 0; nAttribute < knDimension; nAttribute++)
+        {
+            std::cout << cluster.centroid[nAttribute];
+            if(nAttribute != knDimension - 1)
+            {
+                std::cout << ", ";
+            }
+        }
+        std::cout << ")" << std::endl;
+        std::cout << "Number of points: " << cluster.points.size() << std::endl;
+
+        for(std::vector<int>::const_iterator iter = cluster.points.begin();
+            iter != cluster.points.end();
+            iter++)
+        {
+            std::cout << *iter << std::endl;
+        }
+        std::cout << std::endl << std::endl;
+    }
+    double centroid0_x = result.clusters[0].centroid[0];
+    double centroid0_y = result.clusters[0].centroid[1];
+    BOOST_CHECK_EQUAL( centroid0_x, 1.5 );
+    BOOST_CHECK_EQUAL( centroid0_y, 1 );
+
+
+    double centroid1_x = result.clusters[1].centroid[0];
+    double centroid1_y = result.clusters[1].centroid[1];
+    BOOST_CHECK_EQUAL( centroid1_x, 4.5 );
+    BOOST_CHECK_EQUAL( centroid1_y, 3.5 );
+}