From 7c4e0a7c598d823f00fae48b139eb16abef70575 Mon Sep 17 00:00:00 2001
From: Mitch Riedstra <mitch@riedstra.us>
Date: Sat, 9 Mar 2019 10:33:42 -0500
Subject: Complete re-write

---
 stats.go | 93 +++++++++++++++++++++++++++++++++++++++++++++++++++++-----------
 1 file changed, 77 insertions(+), 16 deletions(-)

(limited to 'stats.go')

diff --git a/stats.go b/stats.go
index bc5be7c..ffe94eb 100644
--- a/stats.go
+++ b/stats.go
@@ -1,29 +1,90 @@
 // Provides some basic statisitcs functionality. Specifically tailored around
-// dealing with large streams of data at the moment.
+// dealing with large streams of data
+// It's a Go implimentation of this:
+//
+// https://www.johndcook.com/blog/skewness_kurtosis/
+//
+// Example program:
+//	package main
+//
+//	import (
+//		"git.riedstra.us/go/stats"
+//
+//		"fmt"
+//	)
+//
+//	func main() {
+//		myFakeStreamOfData := []float64{01, 04, 60, 55, 80, 06, 75, 51, 63, 10}
+//		myStats := &stats.Stats{}
+//		for _, entry := range myFakeStreamOfData {
+//			myStats.Push(entry)
+//		}
+//		fmt.Println(myStats.PrettyStats())
+//	}
+// Example output:
+//
+//		Mean: 40.50
+//		Variance: 996.72
+//		Standard Deviation: 31.57
+//		Skewness: -0.20
+//		Kurtosis: -1.66
+//
+//
 package stats
 
 import (
 	"math"
 )
 
-// Stores the average, standard deviation, and the current number of entities
-// processed
+// Structure used to hold the information about the dataset.
+// several methods are exposed in order to make your life easy
 type Stats struct {
-	Mean     float64
-	Variance float64
-	N        float64
+	n, m1, m2, m3, m4 float64
 }
 
-// Returns the square root of 1/2 the variance, or Standard Deviation from the
-// mean
-func (s *Stats) Stdev() float64 {
-	return math.Sqrt(s.Variance / 2)
+// Zeros out the  struct for re-use
+func (s *Stats) Clear() {
+	s.n = 0
+	s.m1, s.m2, s.m3, s.m4 = 0, 0, 0, 0
 }
 
-// Adds an entry to our struct, taking care of adjusting the Mean and Variance
-func (s *Stats) AddEntry(e float64) {
-	m := s.Mean
-	s.Mean += (e - m) / s.N
-	s.Variance += (e - m) * (e - s.Mean)
-	s.N++
+// This function is used to push numbers onto the struct and calculate on the
+// fly the necessary information to output relevant statistics
+func (s *Stats) Push(x float64) {
+	var delta, delta_n, delta_n2, term1, n1 float64
+
+	n1 = s.n
+	s.n++
+	delta = x - s.m1
+	delta_n = delta / s.n
+	delta_n2 = delta_n * delta_n
+	term1 = delta * delta_n * n1
+	s.m1 += delta_n
+	s.m4 += term1*delta_n2*(s.n*s.n-3*s.n+3) + 6*delta_n2*s.m2 - 4*delta_n*s.m3
+	s.m3 += term1*delta_n*(s.n-2) - 3*delta_n*s.m2
+	s.m2 += term1
+}
+
+func (s *Stats) NumValues() float64 {
+	return s.n
+}
+
+func (s *Stats) Mean() float64 {
+	return s.m1
+}
+
+func (s *Stats) Variance() float64 {
+	return s.m2 / (s.n - 1)
+}
+
+func (s *Stats) StandardDeviation() float64 {
+	return math.Sqrt(s.Variance())
+}
+
+func (s *Stats) Skewness() float64 {
+	return math.Sqrt(s.n) * s.m3 / math.Pow(s.m2, 1.5)
+}
+
+func (s *Stats) Kurtosis() float64 {
+	return s.n*s.m4/(s.m2*s.m2) - 3
 }
-- 
cgit v1.2.3