Standard deviation implementation on Measure-Object

src/Microsoft.PowerShell.Commands.Utility/commands/utility/Measure-Object.cs

@@ -34,7 +34,7 @@ public sealed class GenericMeasureInfo : MeasureInfo
         /// </summary>
         public GenericMeasureInfo()
         {
-            Average = Sum = Maximum = Minimum = null;
+            Average = Sum = Maximum = Minimum = StandardDeviation = null;
         }
 
         /// <summary>
@@ -71,6 +71,11 @@ public GenericMeasureInfo()
         ///
         /// </summary>
         public double? Minimum { get; set; }
+
+        /// <summary>
+        /// The Standard Deviation of property values.
+        /// </summary>
+        public double? StandardDeviation { get; set; }
     }
 
     /// <summary>
@@ -90,7 +95,7 @@ public sealed class GenericObjectMeasureInfo : MeasureInfo
         /// </summary>
         public GenericObjectMeasureInfo()
         {
-            Average = Sum = null;
+            Average = Sum = StandardDeviation = null;
             Maximum = Minimum = null;
         }
 
@@ -128,6 +133,11 @@ public GenericObjectMeasureInfo()
         ///
         /// </summary>
         public object Minimum { get; set; }
+
+        /// <summary>
+        /// The Standard Deviation of property values.
+        /// </summary>
+        public double? StandardDeviation { get; set; }
     }
 
     /// <summary>
@@ -225,6 +235,8 @@ private class Statistics
 
             // Generic/Numeric statistics
             internal double sum = 0.0;
+            internal double sumPrevious = 0.0;
+            internal double variance = 0.0;
             internal object max = null;
             internal object min = null;
 
@@ -263,6 +275,24 @@ public MeasureObjectCommand()
 
         #endregion Common parameters in both sets
 
+        /// <summary>
+        /// Set to true if Standard Deviation is to be returned.
+        /// </summary>
+        [Parameter(ParameterSetName = GenericParameterSet)]
+        public SwitchParameter StandardDeviation
+        {
+            get
+            {
+                return _measureStandardDeviation;
+            }
+            set
+            {
+                _measureStandardDeviation = value;
+            }
+        }
+
+        private bool _measureStandardDeviation;
+
         /// <summary>
         /// Set to true is Sum is to be returned
         /// </summary>
@@ -525,7 +555,7 @@ private void AnalyzeValue(string propertyName, object objValue)
                 AnalyzeString(strValue, stat);
             }
 
-            if (_measureAverage || _measureSum)
+            if (_measureAverage || _measureSum || _measureStandardDeviation)
             {
                 double numValue = 0.0;
                 if (!LanguagePrimitives.TryConvertTo(objValue, out numValue))
@@ -708,8 +738,19 @@ private void AnalyzeString(string strValue, Statistics stat)
         /// </summary>
         private void AnalyzeNumber(double numValue, Statistics stat)
         {
-            if (_measureSum || _measureAverage)
+            if (_measureSum || _measureAverage || _measureStandardDeviation)
+            {
+                stat.sumPrevious = stat.sum;
                 stat.sum += numValue;
+            }
+            if (_measureStandardDeviation && stat.count > 1)
+            {
+                // Based off of iterative method of calculating variance on
+                // https://en.wikipedia.org/wiki/Algorithms_for_calculating_variance#Online_algorithm
+                double avgPrevious = stat.sumPrevious / (stat.count - 1);
+                stat.variance *= (stat.count - 2.0) / (stat.count - 1);
+                stat.variance += (numValue - avgPrevious) * (numValue - avgPrevious) / stat.count;
+            }
         }
 
         /// <summary>
@@ -790,15 +831,22 @@ private MeasureInfo CreateGenericMeasureInfo(Statistics stat, bool shouldUseGene
         {
             double? sum = null;
             double? average = null;
+            double? StandardDeviation = null;
             object max = null;
             object min = null;
 
             if (!_nonNumericError)
             {
                 if (_measureSum)
                     sum = stat.sum;
+
                 if (_measureAverage && stat.count > 0)
                     average = stat.sum / stat.count;
+
+                if (_measureStandardDeviation)
+                {
+                    StandardDeviation = Math.Sqrt(stat.variance);
+                }
             }
 
             if (_measureMax)
@@ -835,6 +883,7 @@ private MeasureInfo CreateGenericMeasureInfo(Statistics stat, bool shouldUseGene
                 gmi.Count = stat.count;
                 gmi.Sum = sum;
                 gmi.Average = average;
+                gmi.StandardDeviation = StandardDeviation;
                 if (null != max)
                 {
                     gmi.Maximum = (double)max;
@@ -887,7 +936,7 @@ private TextMeasureInfo CreateTextMeasureInfo(Statistics stat)
 
         /// <summary>
         /// Whether or not a numeric conversion error occurred.
-        /// If true, then average/sum will not be output.
+        /// If true, then average/sum/standard deviation will not be output.
         /// </summary>
         private bool _nonNumericError = false;
 

test/powershell/Modules/Microsoft.PowerShell.Utility/Measure-Object.Tests.ps1

@@ -3,6 +3,7 @@
 Describe "Measure-Object" -Tags "CI" {
     BeforeAll {
         $testObject = 1,3,4
+        $testObject2 = 1..100
     }
 
     It "Should be able to be called without error" {
@@ -17,6 +18,51 @@ Describe "Measure-Object" -Tags "CI" {
         $($testObject | Measure-Object).Count | Should Be $testObject.Length
     }
 
+    It "Should calculate Standard Deviation" {
+        $actual = ($testObject | Measure-Object -StandardDeviation)
+        # We check this way since .StandardDeviation returns a double value
+        # 1.52752523165195 was calculated outside powershell using formula from
+        # http://mathworld.wolfram.com/StandardDeviation.html
+        [Math]::abs($actual.StandardDeviation - 1.52752523165195) | Should -BeLessThan .00000000000001
+    }
+
+
+    It "Should calculate Standard Deviation" {
+        $actual = ($testObject2 | Measure-Object -StandardDeviation)
+        # We check this way since .StandardDeviation returns a double value
+        # 29.011491975882 was calculated outside powershell using formula from
+        # http://mathworld.wolfram.com/StandardDeviation.html
+        [Math]::abs($actual.StandardDeviation - 29.011491975882) | Should -BeLessThan .0000000000001
+    }
+
+    It "Should calculate Standard Deviation with -Sum" {
+        $actual = ($testObject | Measure-Object -Sum -StandardDeviation)
+        # We check this way since .StandardDeviation returns a double value
+        $actual.Sum | Should Be 8
+        # 1.52752523165195 was calculated outside powershell using formula from
+        # http://mathworld.wolfram.com/StandardDeviation.html
+        [Math]::abs($actual.StandardDeviation - 1.52752523165195) | Should -BeLessThan .00000000000001
+    }
+
+    It "Should calculate Standard Deviation with -Average" {
+        $actual = ($testObject | Measure-Object -Average -StandardDeviation)
+        # We check this way since .StandardDeviation returns a double value
+        [Math]::abs($actual.Average - 2.66666666666667) | Should -BeLessThan .00000000000001
+        # 1.52752523165195 was calculated outside powershell using formula from
+        # http://mathworld.wolfram.com/StandardDeviation.html
+        [Math]::abs($actual.StandardDeviation - 1.52752523165195) | Should -BeLessThan .00000000000001
+    }
+
+    It "Should calculate Standard Deviation with -Sum -Average" {
+        $actual = ($testObject2 | Measure-Object -Sum -Average -StandardDeviation)
+        # We check this way since .StandardDeviation returns a double value
+        $actual.Sum | Should Be 5050
+        $actual.Average | Should Be 50.5
+        # 29.011491975882 was calculated outside powershell using formula from
+        # http://mathworld.wolfram.com/StandardDeviation.html
+        [Math]::abs($actual.StandardDeviation - 29.011491975882) | Should -BeLessThan .0000000000001
+    }
+
     It "Should be able to count using the Property switch" {
         $expected = $(Get-ChildItem $TestDrive).Length
         $actual   = $(Get-ChildItem $TestDrive | Measure-Object -Property Length).Count