Mercurial > dive4elements > river

annotate artifacts/src/main/java/org/dive4elements/river/artifacts/math/GrubbsOutlier.java @ 5863:4897a58c8746

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
River artifacts: Added new copyright headers.
author Sascha L. Teichmann <teichmann@intevation.de>
date Sun, 28 Apr 2013 14:40:59 +0200
parents 5aa05a7a34b7
children af13ceeba52a
rev   line source
5863
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
1 /* Copyright (C) 2011, 2012, 2013 by Bundesanstalt für Gewässerkunde
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
2 * Software engineering by Intevation GmbH
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
3 *
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
4 * This file is Free Software under the GNU AGPL (>=v3)
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
5 * and comes with ABSOLUTELY NO WARRANTY! Check out the
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
6 * documentation coming with Dive4Elements River for details.
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
7 */
4897a58c8746 River artifacts: Added new copyright headers.
Sascha L. Teichmann <teichmann@intevation.de>
parents: 5838
diff changeset
8
5831
bd047b71ab37 Repaired internal references
Sascha L. Teichmann <teichmann@intevation.de>
parents: 4794
diff changeset
9 package org.dive4elements.river.artifacts.math;
4794
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
10
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
11 import java.util.List;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
12
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
13 import org.apache.commons.math.MathException;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
14
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
15 import org.apache.commons.math.distribution.TDistributionImpl;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
16
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
17 import org.apache.commons.math.stat.descriptive.moment.Mean;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
18 import org.apache.commons.math.stat.descriptive.moment.StandardDeviation;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
19
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
20 import org.apache.log4j.Logger;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
21
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
22 public class GrubbsOutlier
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
23 {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
24 public static final double EPSILON = 1e-5;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
25
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
26 public static final double DEFAULT_ALPHA = 0.05;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
27
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
28 private static Logger log = Logger.getLogger(GrubbsOutlier.class);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
29
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
30 protected GrubbsOutlier() {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
31 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
32
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
33 public static Integer findOutlier(List<Double> values) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
34 return findOutlier(values, DEFAULT_ALPHA, null);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
35 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
36
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
37 public static Integer findOutlier(
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
38 List<Double> values,
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
39 double alpha,
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
40 double[] stdDevResult
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
41 ) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
42 boolean debug = log.isDebugEnabled();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
43
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
44 if (debug) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
45 log.debug("outliers significance: " + alpha);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
46 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
47
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
48 alpha = 1d - alpha;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
49
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
50 int N = values.size();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
51
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
52 if (debug) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
53 log.debug("Values to check: " + N);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
54 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
55
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
56 if (N < 3) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
57 return null;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
58 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
59
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
60 Mean mean = new Mean();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
61 StandardDeviation std = new StandardDeviation();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
62
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
63 for (Double value: values) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
64 double v = value.doubleValue();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
65 mean.increment(v);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
66 std .increment(v);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
67 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
68
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
69 double m = mean.getResult();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
70 double s = std.getResult();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
71
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
72 if (debug) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
73 log.debug("mean: " + m);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
74 log.debug("std dev: " + s);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
75 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
76
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
77 double maxZ = -Double.MAX_VALUE;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
78 int iv = -1;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
79 for (int i = N-1; i >= 0; --i) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
80 double v = values.get(i).doubleValue();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
81 double z = Math.abs(v - m);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
82 if (z > maxZ) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
83 maxZ = z;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
84 iv = i;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
85 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
86 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
87
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
88 if (Math.abs(s) < EPSILON) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
89 return null;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
90 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
91
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
92 maxZ /= s;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
93
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
94 TDistributionImpl tdist = new TDistributionImpl(N-2);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
95
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
96 double t;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
97
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
98 try {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
99 t = tdist.inverseCumulativeProbability(alpha/(N+N));
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
100 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
101 catch (MathException me) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
102 log.error(me);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
103 return null;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
104 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
105
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
106 t *= t;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
107
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
108 double za = ((N-1)/Math.sqrt(N))*Math.sqrt(t/(N-2d+t));
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
109
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
110 if (debug) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
111 log.debug("max: " + maxZ + " crit: " + za);
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
112 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
113 if (stdDevResult != null) {
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
114 stdDevResult[0] = std.getResult();
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
115 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
116 return maxZ > za
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
117 ? Integer.valueOf(iv)
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
118 : null;
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
119 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
120 }
a7d080347ac3 MINFO: Allow two methods for outlier test in SQ relation.
Raimund Renkert <rrenkert@intevation.de>
parents:
diff changeset
121 // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 :
http://dive4elements.wald.intevation.org