From f31809a8cddd7444bafc593482845aa4226b3e87 Mon Sep 17 00:00:00 2001
From: Nick Hamblet <nhamblet@ccri.com>
Date: Mon, 6 Oct 2014 11:14:31 -0400
Subject: [PATCH] Added a GaussianComparator, with unit tests

This seems like a good replacement for NumericComparator, which has issues with negative
values (issue #181), and also doesn't take scale into account (e.g., 0.01 may be far from
0.0 in some instances, but effectively the same in others).
---
 .../duke/comparators/GaussianComparator.java  | 46 +++++++++++++++++++
 .../duke/test/GaussianComparatorTest.java     | 44 ++++++++++++++++++
 2 files changed, 90 insertions(+)
 create mode 100644 src/main/java/no/priv/garshol/duke/comparators/GaussianComparator.java
 create mode 100644 src/test/java/no/priv/garshol/duke/test/GaussianComparatorTest.java

diff --git a/src/main/java/no/priv/garshol/duke/comparators/GaussianComparator.java b/src/main/java/no/priv/garshol/duke/comparators/GaussianComparator.java
new file mode 100644
index 00000000..a39392fe
--- /dev/null
+++ b/src/main/java/no/priv/garshol/duke/comparators/GaussianComparator.java
@@ -0,0 +1,46 @@
+package no.priv.garshol.duke.comparators;
+
+import no.priv.garshol.duke.Comparator;
+
+public class GaussianComparator implements Comparator {
+
+    private Double sigma = 1.0;
+    private Double sigmaSq = 1.0;
+    private Double powerCoef = -0.5;
+    private final Double sqrtTwoPi = 2.50662827;
+    private Double linearCoef = 1.0 / sqrtTwoPi;
+
+    public boolean isTokenized() {
+        return false;
+    }
+
+    public void setSigma(Double sigma) {
+        this.sigma = sigma;
+        updateValues();
+    }
+
+    // We return the probability (almost) of drawing v1 from a gaussian with mean v2 and standard deviation sigma.
+    // It's not the honest probability, because we don't multiply by the normalizing factor (linearCoef).
+    // This makes values somewhat easier to think about, because you don't have to know the maximum value of the
+    // normal distribution for your chosen sigma.
+    public double compare(String v1, String v2) {
+        // begin like NumericComparator
+        double d1 = 0.0;
+        double d2 = 0.0;
+        try {
+            d1 = Double.parseDouble(v1);
+            d2 = Double.parseDouble(v2);
+        } catch (NumberFormatException e) {
+            return 0.5; // we just ignore this. whether it's wise I'm not sure
+        }
+
+        double diff = Math.abs(d1 - d2);
+        return Math.exp(powerCoef * diff * diff);
+    }
+
+    private void updateValues() {
+        sigmaSq = sigma * sigma;
+        powerCoef = -1.0 / (2.0 * sigmaSq);
+        linearCoef = 1.0 / (sigma * sqrtTwoPi);
+    }
+}
diff --git a/src/test/java/no/priv/garshol/duke/test/GaussianComparatorTest.java b/src/test/java/no/priv/garshol/duke/test/GaussianComparatorTest.java
new file mode 100644
index 00000000..636529a3
--- /dev/null
+++ b/src/test/java/no/priv/garshol/duke/test/GaussianComparatorTest.java
@@ -0,0 +1,44 @@
+package no.priv.garshol.duke.test;
+
+import org.junit.Test;
+import org.junit.Before;
+import static junit.framework.Assert.assertEquals;
+
+import no.priv.garshol.duke.comparators.GaussianComparator;
+
+public class GaussianComparatorTest {
+
+    private GaussianComparator defaultComparator;
+    private GaussianComparator sigma10Comparator;
+
+    @Before
+    public void setUp() {
+        defaultComparator = new GaussianComparator();
+        sigma10Comparator = new GaussianComparator();
+        sigma10Comparator.setSigma(10.0);
+    }
+
+    @Test
+    public void testEqual() {
+        assertEquals(1.0, defaultComparator.compare("42", "42"));
+        assertEquals(1.0, sigma10Comparator.compare("42", "42"));
+    }
+
+    @Test
+    public void testEqual2() {
+        assertEquals(1.0, defaultComparator.compare("42.0", "42.0"));
+        assertEquals(1.0, sigma10Comparator.compare("42.0", "42.0"));
+    }
+
+    @Test
+    public void testClose() {
+        assertEquals(0.1353, defaultComparator.compare("40", "42"), 0.0001);
+        assertEquals(0.9801, sigma10Comparator.compare("40", "42"), 0.0001);
+    }
+
+    @Test
+    public void testFar() {
+        assertEquals(0.0, defaultComparator.compare("25", "42"), 0.0001);
+        assertEquals(0.2357, sigma10Comparator.compare("25", "42"), 0.0001);
+    }
+}