From 69a8b656da5513f125c645b325ee7a5f9f7c1394 Mon Sep 17 00:00:00 2001
From: Christoph Ruegg <git@cdrnet.ch>
Date: Wed, 16 Apr 2014 01:42:21 +0200
Subject: [PATCH] Statistics: DescriptiveStatistics must use a long counter to
 support large sample sets

---
 .../Statistics/DescriptiveStatistics.cs       | 12 +++++-----
 .../StatisticsTests/StatisticsTests.cs        | 23 +++++++++++++++++++
 2 files changed, 29 insertions(+), 6 deletions(-)

diff --git a/src/Numerics/Statistics/DescriptiveStatistics.cs b/src/Numerics/Statistics/DescriptiveStatistics.cs
index ccb97c6c..e49c8444 100644
--- a/src/Numerics/Statistics/DescriptiveStatistics.cs
+++ b/src/Numerics/Statistics/DescriptiveStatistics.cs
@@ -104,7 +104,7 @@ namespace MathNet.Numerics.Statistics
         /// Gets the size of the sample.
         /// </summary>
         /// <value>The size of the sample.</value>
-        public int Count { get; private set; }
+        public long Count { get; private set; }
 
         /// <summary>
         /// Gets the sample mean.
@@ -162,7 +162,7 @@ namespace MathNet.Numerics.Statistics
             double kurtosis = 0;
             double minimum = Double.PositiveInfinity;
             double maximum = Double.NegativeInfinity;
-            int n = 0;
+            long n = 0;
             foreach (var xi in data)
             {
                 double delta = xi - mean;
@@ -202,7 +202,7 @@ namespace MathNet.Numerics.Statistics
             double kurtosis = 0;
             double minimum = Double.PositiveInfinity;
             double maximum = Double.NegativeInfinity;
-            int n = 0;
+            long n = 0;
             foreach (var xi in data)
             {
                 if (xi.HasValue)
@@ -244,7 +244,7 @@ namespace MathNet.Numerics.Statistics
             decimal kurtosis = 0;
             decimal minimum = Decimal.MaxValue;
             decimal maximum = Decimal.MinValue;
-            int n = 0;
+            long n = 0;
             foreach (double x in data)
             {
                 decimal xi = (decimal)x;
@@ -284,7 +284,7 @@ namespace MathNet.Numerics.Statistics
             decimal kurtosis = 0;
             decimal minimum = Decimal.MaxValue;
             decimal maximum = Decimal.MinValue;
-            int n = 0;
+            long n = 0;
             foreach (double? x in data)
             {
                 if (x.HasValue)
@@ -325,7 +325,7 @@ namespace MathNet.Numerics.Statistics
         /// <param name="minimum">For setting Minimum.</param>
         /// <param name="maximum">For setting Maximum.</param>
         /// <param name="n">For setting Count.</param>
-        void SetStatistics(double mean, double variance, double skewness, double kurtosis, double minimum, double maximum, int n)
+        void SetStatistics(double mean, double variance, double skewness, double kurtosis, double minimum, double maximum, long n)
         {
             Mean = mean;
             Count = n;
diff --git a/src/UnitTests/StatisticsTests/StatisticsTests.cs b/src/UnitTests/StatisticsTests/StatisticsTests.cs
index 8c84c349..87a16ece 100644
--- a/src/UnitTests/StatisticsTests/StatisticsTests.cs
+++ b/src/UnitTests/StatisticsTests/StatisticsTests.cs
@@ -988,6 +988,29 @@ namespace MathNet.Numerics.UnitTests.StatisticsTests
             Assert.That(SortedArrayStatistics.Median(new[] { double.NegativeInfinity, 2.0, double.PositiveInfinity }), Is.EqualTo(2.0));
             Assert.That(SortedArrayStatistics.Median(new[] { double.NegativeInfinity, 2.0, 3.0, double.PositiveInfinity }), Is.EqualTo(2.5));
         }
+
+        [Test]
+        public void RobustOnLargeSampleSets()
+        {
+            // 0, 0.25, 0.5, 0.75, 0, 0.25, 0.5, 0.75, ...
+            var shorter = Generate.Periodic(4*4096, 4, 1);
+            var longer = Generate.Periodic(4*32768, 4, 1);
+
+            Assert.That(Statistics.Mean(shorter), Is.EqualTo(0.375).Within(1e-14), "Statistics.Mean: shorter");
+            Assert.That(Statistics.Mean(longer), Is.EqualTo(0.375).Within(1e-14), "Statistics.Mean: longer");
+            Assert.That(new DescriptiveStatistics(shorter).Mean, Is.EqualTo(0.375).Within(1e-14), "DescriptiveStatistics.Mean: shorter");
+            Assert.That(new DescriptiveStatistics(longer).Mean, Is.EqualTo(00.375).Within(1e-14), "DescriptiveStatistics.Mean: longer");
+
+            Assert.That(Statistics.Skewness(shorter), Is.EqualTo(0.0).Within(1e-12), "Statistics.Skewness: shorter");
+            Assert.That(Statistics.Skewness(longer), Is.EqualTo(0.0).Within(1e-12), "Statistics.Skewness: longer");
+            Assert.That(new DescriptiveStatistics(shorter).Skewness, Is.EqualTo(0.0).Within(1e-12), "DescriptiveStatistics.Skewness: shorter");
+            Assert.That(new DescriptiveStatistics(longer).Skewness, Is.EqualTo(0.0).Within(1e-12), "DescriptiveStatistics.Skewness: longer");
+
+            Assert.That(Statistics.Kurtosis(shorter), Is.EqualTo(-1.36).Within(1e-4), "Statistics.Kurtosis: shorter");
+            Assert.That(Statistics.Kurtosis(longer), Is.EqualTo(-1.36).Within(1e-4), "Statistics.Kurtosis: longer");
+            Assert.That(new DescriptiveStatistics(shorter).Kurtosis, Is.EqualTo(-1.36).Within(1e-4), "DescriptiveStatistics.Kurtosis: shorter");
+            Assert.That(new DescriptiveStatistics(longer).Kurtosis, Is.EqualTo(-1.36).Within(1e-4), "DescriptiveStatistics.Kurtosis: longer");
+        }
     }
 }