1 | #region License Information
|
---|
2 | /* HeuristicLab
|
---|
3 | * Copyright (C) 2002-2015 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
|
---|
4 | *
|
---|
5 | * This file is part of HeuristicLab.
|
---|
6 | *
|
---|
7 | * HeuristicLab is free software: you can redistribute it and/or modify
|
---|
8 | * it under the terms of the GNU General Public License as published by
|
---|
9 | * the Free Software Foundation, either version 3 of the License, or
|
---|
10 | * (at your option) any later version.
|
---|
11 | *
|
---|
12 | * HeuristicLab is distributed in the hope that it will be useful,
|
---|
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
15 | * GNU General Public License for more details.
|
---|
16 | *
|
---|
17 | * You should have received a copy of the GNU General Public License
|
---|
18 | * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
|
---|
19 | */
|
---|
20 | #endregion
|
---|
21 |
|
---|
22 | using System;
|
---|
23 | using System.Collections.Generic;
|
---|
24 | using HeuristicLab.Common;
|
---|
25 |
|
---|
26 | namespace HeuristicLab.Analysis.Statistics {
|
---|
27 | public static class KruskalWallisTest {
|
---|
28 | /// <summary>
|
---|
29 | /// Performs the Kruskal-Wallis test and returns the p-Value.
|
---|
30 | /// (source based on R's kruskal.test(), GNU GPL)
|
---|
31 | /// </summary>
|
---|
32 | public static double Test(double[][] data) {
|
---|
33 | double[] g;
|
---|
34 | double[] x = FlattenArray(data, out g);
|
---|
35 | int n = x.Length;
|
---|
36 | int parameter = data.Length - 1;
|
---|
37 |
|
---|
38 | int[] r = Rank(x);
|
---|
39 | double[][] ties = CountDuplicates(x);
|
---|
40 | double statistic = CalculateStatistic(r, g);
|
---|
41 | double tiesCorrection = CalculateTiesCorrection(ties);
|
---|
42 | statistic = ((12 * statistic / (n * (n + 1)) - 3 * (n + 1)) /
|
---|
43 | (1 - tiesCorrection / (Math.Pow(n, 3) - n)));
|
---|
44 |
|
---|
45 | return alglib.chisquarecdistribution(parameter, statistic);
|
---|
46 | }
|
---|
47 |
|
---|
48 | private static double CalculateStatistic(int[] r, double[] g) {
|
---|
49 | double result = 0.0;
|
---|
50 | double lastG = g[0];
|
---|
51 | double curSum = 0.0;
|
---|
52 | int cnt = 0;
|
---|
53 | for (int i = 0; i < r.Length; i++) {
|
---|
54 | if (lastG.IsAlmost(g[i])) {
|
---|
55 | curSum += r[i];
|
---|
56 | cnt++;
|
---|
57 | } else {
|
---|
58 | double sum = Math.Pow(curSum, 2);
|
---|
59 | sum /= cnt;
|
---|
60 | result += sum;
|
---|
61 |
|
---|
62 | lastG = g[i];
|
---|
63 | curSum = r[i];
|
---|
64 | cnt = 1;
|
---|
65 | }
|
---|
66 | }
|
---|
67 |
|
---|
68 | double lastSum = Math.Pow(curSum, 2);
|
---|
69 | lastSum /= cnt;
|
---|
70 | result += lastSum;
|
---|
71 |
|
---|
72 | return result;
|
---|
73 | }
|
---|
74 |
|
---|
75 | private static double CalculateTiesCorrection(double[][] ties) {
|
---|
76 | double sum = 0.0;
|
---|
77 | for (int i = 0; i < ties[1].Length; i++) {
|
---|
78 | double tic = Math.Pow(ties[1][i], 3) - ties[1][i];
|
---|
79 | sum += tic;
|
---|
80 | }
|
---|
81 | return sum;
|
---|
82 | }
|
---|
83 |
|
---|
84 | private static double[] FlattenArray(double[][] x, out double[] indizes) {
|
---|
85 | int compLenght = 0;
|
---|
86 | for (int i = 0; i < x.Length; i++) {
|
---|
87 | compLenght += x[i].Length;
|
---|
88 | }
|
---|
89 |
|
---|
90 | double[] result = new double[compLenght];
|
---|
91 | indizes = new double[compLenght];
|
---|
92 |
|
---|
93 | int resultPos = 0;
|
---|
94 | for (int i = 0; i < x.Length; i++) {
|
---|
95 | Array.Copy(x[i], 0, result, resultPos, x[i].Length);
|
---|
96 |
|
---|
97 | for (int j = resultPos; j < resultPos + x[i].Length; j++) {
|
---|
98 | indizes[j] = i;
|
---|
99 | }
|
---|
100 | resultPos += x[i].Length;
|
---|
101 | }
|
---|
102 |
|
---|
103 | return result;
|
---|
104 | }
|
---|
105 |
|
---|
106 | private static double[][] CountDuplicates(double[] x) {
|
---|
107 | List<double> number = new List<double>();
|
---|
108 | List<double> cnt = new List<double>();
|
---|
109 | double[] sortedX = new double[x.Length];
|
---|
110 |
|
---|
111 | Array.Copy(x, sortedX, x.Length);
|
---|
112 | Array.Sort(sortedX);
|
---|
113 |
|
---|
114 | double last = x[0];
|
---|
115 | number.Add(x[0]);
|
---|
116 | cnt.Add(1);
|
---|
117 |
|
---|
118 | for (int i = 1; i < x.Length; i++) {
|
---|
119 | if (!last.IsAlmost(x[i])) {
|
---|
120 | number.Add(x[i]);
|
---|
121 | last = x[i];
|
---|
122 | cnt.Add(1);
|
---|
123 | } else {
|
---|
124 | cnt[cnt.Count - 1] += 1;
|
---|
125 | }
|
---|
126 | }
|
---|
127 |
|
---|
128 | double[][] result = new double[2][];
|
---|
129 | result[0] = number.ToArray();
|
---|
130 | result[1] = cnt.ToArray();
|
---|
131 |
|
---|
132 | return result;
|
---|
133 | }
|
---|
134 |
|
---|
135 | private static int[] Rank(double[] x) {
|
---|
136 | double[] keys = new double[x.Length];
|
---|
137 | int[] items = new int[x.Length];
|
---|
138 | int[] ranks = new int[x.Length];
|
---|
139 |
|
---|
140 | Array.Copy(x, keys, x.Length);
|
---|
141 | for (int i = 0; i < x.Length; i++) items[i] = i;
|
---|
142 |
|
---|
143 | Array.Sort(keys, items);
|
---|
144 |
|
---|
145 | for (int i = 0; i < x.Length; i++) {
|
---|
146 | ranks[items[i]] = i + 1;
|
---|
147 | }
|
---|
148 | return ranks;
|
---|
149 | }
|
---|
150 | }
|
---|
151 | }
|
---|