1 | /*************************************************************************
|
---|
2 | Copyright (c) 2007, Sergey Bochkanov (ALGLIB project).
|
---|
3 |
|
---|
4 | >>> SOURCE LICENSE >>>
|
---|
5 | This program is free software; you can redistribute it and/or modify
|
---|
6 | it under the terms of the GNU General Public License as published by
|
---|
7 | the Free Software Foundation (www.fsf.org); either version 2 of the
|
---|
8 | License, or (at your option) any later version.
|
---|
9 |
|
---|
10 | This program is distributed in the hope that it will be useful,
|
---|
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
13 | GNU General Public License for more details.
|
---|
14 |
|
---|
15 | A copy of the GNU General Public License is available at
|
---|
16 | http://www.fsf.org/licensing/licenses
|
---|
17 |
|
---|
18 | >>> END OF LICENSE >>>
|
---|
19 | *************************************************************************/
|
---|
20 |
|
---|
21 | using System;
|
---|
22 |
|
---|
23 | namespace alglib
|
---|
24 | {
|
---|
25 | public class correlation
|
---|
26 | {
|
---|
27 | /*************************************************************************
|
---|
28 | Pearson product-moment correlation coefficient
|
---|
29 |
|
---|
30 | Input parameters:
|
---|
31 | X - sample 1 (array indexes: [0..N-1])
|
---|
32 | Y - sample 2 (array indexes: [0..N-1])
|
---|
33 | N - sample size.
|
---|
34 |
|
---|
35 | Result:
|
---|
36 | Pearson product-moment correlation coefficient
|
---|
37 |
|
---|
38 | -- ALGLIB --
|
---|
39 | Copyright 09.04.2007 by Bochkanov Sergey
|
---|
40 | *************************************************************************/
|
---|
41 | public static double pearsoncorrelation(ref double[] x,
|
---|
42 | ref double[] y,
|
---|
43 | int n)
|
---|
44 | {
|
---|
45 | double result = 0;
|
---|
46 | int i = 0;
|
---|
47 | double xmean = 0;
|
---|
48 | double ymean = 0;
|
---|
49 | double s = 0;
|
---|
50 | double xv = 0;
|
---|
51 | double yv = 0;
|
---|
52 | double t1 = 0;
|
---|
53 | double t2 = 0;
|
---|
54 |
|
---|
55 | xv = 0;
|
---|
56 | yv = 0;
|
---|
57 | if( n<=1 )
|
---|
58 | {
|
---|
59 | result = 0;
|
---|
60 | return result;
|
---|
61 | }
|
---|
62 |
|
---|
63 | //
|
---|
64 | // Mean
|
---|
65 | //
|
---|
66 | xmean = 0;
|
---|
67 | ymean = 0;
|
---|
68 | for(i=0; i<=n-1; i++)
|
---|
69 | {
|
---|
70 | xmean = xmean+x[i];
|
---|
71 | ymean = ymean+y[i];
|
---|
72 | }
|
---|
73 | xmean = xmean/n;
|
---|
74 | ymean = ymean/n;
|
---|
75 |
|
---|
76 | //
|
---|
77 | // numerator and denominator
|
---|
78 | //
|
---|
79 | s = 0;
|
---|
80 | xv = 0;
|
---|
81 | yv = 0;
|
---|
82 | for(i=0; i<=n-1; i++)
|
---|
83 | {
|
---|
84 | t1 = x[i]-xmean;
|
---|
85 | t2 = y[i]-ymean;
|
---|
86 | xv = xv+AP.Math.Sqr(t1);
|
---|
87 | yv = yv+AP.Math.Sqr(t2);
|
---|
88 | s = s+t1*t2;
|
---|
89 | }
|
---|
90 | if( xv==0 | yv==0 )
|
---|
91 | {
|
---|
92 | result = 0;
|
---|
93 | }
|
---|
94 | else
|
---|
95 | {
|
---|
96 | result = s/(Math.Sqrt(xv)*Math.Sqrt(yv));
|
---|
97 | }
|
---|
98 | return result;
|
---|
99 | }
|
---|
100 |
|
---|
101 |
|
---|
102 | /*************************************************************************
|
---|
103 | Spearman's rank correlation coefficient
|
---|
104 |
|
---|
105 | Input parameters:
|
---|
106 | X - sample 1 (array indexes: [0..N-1])
|
---|
107 | Y - sample 2 (array indexes: [0..N-1])
|
---|
108 | N - sample size.
|
---|
109 |
|
---|
110 | Result:
|
---|
111 | Spearman's rank correlation coefficient
|
---|
112 |
|
---|
113 | -- ALGLIB --
|
---|
114 | Copyright 09.04.2007 by Bochkanov Sergey
|
---|
115 | *************************************************************************/
|
---|
116 | public static double spearmanrankcorrelation(double[] x,
|
---|
117 | double[] y,
|
---|
118 | int n)
|
---|
119 | {
|
---|
120 | double result = 0;
|
---|
121 |
|
---|
122 | x = (double[])x.Clone();
|
---|
123 | y = (double[])y.Clone();
|
---|
124 |
|
---|
125 | rankx(ref x, n);
|
---|
126 | rankx(ref y, n);
|
---|
127 | result = pearsoncorrelation(ref x, ref y, n);
|
---|
128 | return result;
|
---|
129 | }
|
---|
130 |
|
---|
131 |
|
---|
132 | /*************************************************************************
|
---|
133 | Internal ranking subroutine
|
---|
134 | *************************************************************************/
|
---|
135 | private static void rankx(ref double[] x,
|
---|
136 | int n)
|
---|
137 | {
|
---|
138 | int i = 0;
|
---|
139 | int j = 0;
|
---|
140 | int k = 0;
|
---|
141 | int t = 0;
|
---|
142 | double tmp = 0;
|
---|
143 | int tmpi = 0;
|
---|
144 | double[] r = new double[0];
|
---|
145 | int[] c = new int[0];
|
---|
146 |
|
---|
147 |
|
---|
148 | //
|
---|
149 | // Prepare
|
---|
150 | //
|
---|
151 | if( n<=1 )
|
---|
152 | {
|
---|
153 | x[0] = 1;
|
---|
154 | return;
|
---|
155 | }
|
---|
156 | r = new double[n-1+1];
|
---|
157 | c = new int[n-1+1];
|
---|
158 | for(i=0; i<=n-1; i++)
|
---|
159 | {
|
---|
160 | r[i] = x[i];
|
---|
161 | c[i] = i;
|
---|
162 | }
|
---|
163 |
|
---|
164 | //
|
---|
165 | // sort {R, C}
|
---|
166 | //
|
---|
167 | if( n!=1 )
|
---|
168 | {
|
---|
169 | i = 2;
|
---|
170 | do
|
---|
171 | {
|
---|
172 | t = i;
|
---|
173 | while( t!=1 )
|
---|
174 | {
|
---|
175 | k = t/2;
|
---|
176 | if( r[k-1]>=r[t-1] )
|
---|
177 | {
|
---|
178 | t = 1;
|
---|
179 | }
|
---|
180 | else
|
---|
181 | {
|
---|
182 | tmp = r[k-1];
|
---|
183 | r[k-1] = r[t-1];
|
---|
184 | r[t-1] = tmp;
|
---|
185 | tmpi = c[k-1];
|
---|
186 | c[k-1] = c[t-1];
|
---|
187 | c[t-1] = tmpi;
|
---|
188 | t = k;
|
---|
189 | }
|
---|
190 | }
|
---|
191 | i = i+1;
|
---|
192 | }
|
---|
193 | while( i<=n );
|
---|
194 | i = n-1;
|
---|
195 | do
|
---|
196 | {
|
---|
197 | tmp = r[i];
|
---|
198 | r[i] = r[0];
|
---|
199 | r[0] = tmp;
|
---|
200 | tmpi = c[i];
|
---|
201 | c[i] = c[0];
|
---|
202 | c[0] = tmpi;
|
---|
203 | t = 1;
|
---|
204 | while( t!=0 )
|
---|
205 | {
|
---|
206 | k = 2*t;
|
---|
207 | if( k>i )
|
---|
208 | {
|
---|
209 | t = 0;
|
---|
210 | }
|
---|
211 | else
|
---|
212 | {
|
---|
213 | if( k<i )
|
---|
214 | {
|
---|
215 | if( r[k]>r[k-1] )
|
---|
216 | {
|
---|
217 | k = k+1;
|
---|
218 | }
|
---|
219 | }
|
---|
220 | if( r[t-1]>=r[k-1] )
|
---|
221 | {
|
---|
222 | t = 0;
|
---|
223 | }
|
---|
224 | else
|
---|
225 | {
|
---|
226 | tmp = r[k-1];
|
---|
227 | r[k-1] = r[t-1];
|
---|
228 | r[t-1] = tmp;
|
---|
229 | tmpi = c[k-1];
|
---|
230 | c[k-1] = c[t-1];
|
---|
231 | c[t-1] = tmpi;
|
---|
232 | t = k;
|
---|
233 | }
|
---|
234 | }
|
---|
235 | }
|
---|
236 | i = i-1;
|
---|
237 | }
|
---|
238 | while( i>=1 );
|
---|
239 | }
|
---|
240 |
|
---|
241 | //
|
---|
242 | // compute tied ranks
|
---|
243 | //
|
---|
244 | i = 0;
|
---|
245 | while( i<=n-1 )
|
---|
246 | {
|
---|
247 | j = i+1;
|
---|
248 | while( j<=n-1 )
|
---|
249 | {
|
---|
250 | if( r[j]!=r[i] )
|
---|
251 | {
|
---|
252 | break;
|
---|
253 | }
|
---|
254 | j = j+1;
|
---|
255 | }
|
---|
256 | for(k=i; k<=j-1; k++)
|
---|
257 | {
|
---|
258 | r[k] = 1+((double)(i+j-1))/(double)(2);
|
---|
259 | }
|
---|
260 | i = j;
|
---|
261 | }
|
---|
262 |
|
---|
263 | //
|
---|
264 | // back to x
|
---|
265 | //
|
---|
266 | for(i=0; i<=n-1; i++)
|
---|
267 | {
|
---|
268 | x[c[i]] = r[i];
|
---|
269 | }
|
---|
270 | }
|
---|
271 | }
|
---|
272 | }
|
---|