1 files changed, 141 insertions, 0 deletions
diff --git a/src/de/lmu/ifi/dbs/elki/math/statistics/dependence/CorrelationDependenceMeasure.java b/src/de/lmu/ifi/dbs/elki/math/statistics/dependence/CorrelationDependenceMeasure.java
new file mode 100644
index 00000000..e03601f7
--- /dev/null
+++ b/src/de/lmu/ifi/dbs/elki/math/statistics/dependence/CorrelationDependenceMeasure.java
@@ -0,0 +1,141 @@
+package de.lmu.ifi.dbs.elki.math.statistics.dependence;
+
+/*
+ This file is part of ELKI:
+ Environment for Developing KDD-Applications Supported by Index-Structures
+
+ Copyright (C) 2014
+ Ludwig-Maximilians-Universität München
+ Lehr- und Forschungseinheit für Datenbanksysteme
+ ELKI Development Team
+
+ This program is free software: you can redistribute it and/or modify
+ it under the terms of the GNU Affero General Public License as published by
+ the Free Software Foundation, either version 3 of the License, or
+ (at your option) any later version.
+
+ This program is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ GNU Affero General Public License for more details.
+
+ You should have received a copy of the GNU Affero General Public License
+ along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+import java.util.List;
+
+import de.lmu.ifi.dbs.elki.logging.Logging;
+import de.lmu.ifi.dbs.elki.utilities.datastructures.arraylike.NumberArrayAdapter;
+import de.lmu.ifi.dbs.elki.utilities.optionhandling.AbstractParameterizer;
+
+/**
+ * Pearson product-moment correlation coefficient.
+ * 
+ * @author Erich Schubert
+ */
+public class CorrelationDependenceMeasure extends AbstractDependenceMeasure {
+  /**
+   * Class logger.
+   */
+  private static final Logging LOG = Logging.getLogger(CorrelationDependenceMeasure.class);
+
+  /**
+   * Static instance.
+   */
+  public static final CorrelationDependenceMeasure STATIC = new CorrelationDependenceMeasure();
+
+  /**
+   * Constructor - use {@link #STATIC} instance.
+   */
+  protected CorrelationDependenceMeasure() {
+    super();
+  }
+
+  @Override
+  public <A, B> double dependence(NumberArrayAdapter<?, A> adapter1, A data1, NumberArrayAdapter<?, B> adapter2, B data2) {
+    final int len = size(adapter1, data1, adapter2, data2);
+    // Perform two-pass estimation, which is numerically stable and often faster
+    // than the Knuth-Welford approach (see PearsonCorrelation class)
+    double m1 = 0., m2 = 0;
+    for(int i = 0; i < len; i++) {
+      m1 += adapter1.getDouble(data1, i);
+      m2 += adapter2.getDouble(data2, i);
+    }
+    m1 /= len;
+    m2 /= len;
+    // Second pass: variances and covariance
+    double v1 = 0., v2 = 0., cov = 0.;
+    for(int i = 0; i < len; i++) {
+      double d1 = adapter1.getDouble(data1, i) - m1;
+      double d2 = adapter2.getDouble(data2, i) - m2;
+      v1 += d1 * d1;
+      v2 += d2 * d2;
+      cov += d1 * d2;
+    }
+    // Note: we did not normalize by len, as this cancels out.
+    return cov / Math.sqrt(v1 * v2);
+  }
+
+  @Override
+  public <A> double[] dependence(NumberArrayAdapter<?, A> adapter, List<? extends A> data) {
+    final int dims = data.size();
+    final int len = size(adapter, data);
+    double[] means = new double[dims];
+    // Two passes - often faster due to the lower numerical cost
+    // And accurate, don't use sum-of-squares.
+    for(int j = 0; j < dims; j++) {
+      double m = 0.;
+      A da = data.get(j);
+      for(int i = 0; i < len; i++) {
+        m += adapter.getDouble(da, i);
+      }
+      means[j] = m / len;
+    }
+    // Build the covariance matrix, lower triangular half
+    double[] vst = new double[dims];
+    double[] cov = new double[(dims * (dims - 1)) >> 1];
+    double[] buf = new double[dims];
+    for(int i = 0; i < len; i++) {
+      for(int j = 0; j < dims; j++) {
+        buf[j] = adapter.getDouble(data.get(j), i) - means[j];
+      }
+      for(int y = 0, c = 0; y < dims; y++) {
+        for(int x = 0; x < y; x++) {
+          cov[c++] += buf[x] * buf[y];
+        }
+        vst[y] += buf[y] * buf[y];
+      }
+    }
+    // Compute standard deviations (times sqrt(len)!):
+    for(int y = 0; y < dims; y++) {
+      if(vst[y] == 0.) {
+        LOG.warning("Correlation is not well defined for constant attributes.");
+      }
+      vst[y] = Math.sqrt(vst[y]);
+    }
+    for(int y = 1, c = 0; y < dims; y++) {
+      for(int x = 0; x < y; x++) {
+        // We don't need to divide by sqrt(len), because it will cancel out with
+        // the division we skipped just above.
+        cov[c] = cov[c] / (vst[x] * vst[y]);
+        c++;
+      }
+    }
+    return cov;
+  }
+
+  /**
+   * Parameterization class
+   * 
+   * @author Erich Schubert
+   *
+   * @apiviz.exclude
+   */
+  public static class Parameterizer extends AbstractParameterizer {
+    @Override
+    protected CorrelationDependenceMeasure makeInstance() {
+      return STATIC;
+    }
+  }
+}