1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
| /*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
/*
* SelfOrganizingMap.java
* Copyright (C) 2000-2010 University of Waikato, Hamilton, New Zealand
*
*/
package weka.clusterers;
import java.util.Enumeration;
import java.util.Vector;
import weka.core.Attribute;
import weka.core.Capabilities;
import weka.core.Capabilities.Capability;
import weka.core.DenseInstance;
import weka.core.Instance;
import weka.core.Instances;
import weka.core.Option;
import weka.core.OptionHandler;
import weka.core.RevisionUtils;
import weka.core.Utils;
import weka.core.EuclideanDistance;
/**
<!-- globalinfo-start -->
* A Clusterer that implements Kohonen's Self-Organizing Map algorithm for
* unsupervised clustering. <br/>
* See T. Kohonen, Self-Organization and Associative Memory, 3rd Edition, Springer, 1989
* <p/>
<!-- globalinfo-end -->
*
<!-- options-start -->
* Valid options are: <p/>
*
* <pre> -L < initial learning rate>
* The initial learning rate for the training algorithm.
* (Value should be greater than 0.01 and less or equal to 1, Default = 1).</pre>
*
* <pre> -O <number of epochs in ordering phase>
* Number of epochs in ordering phase.
* (Value should be greater than or equal to 2000, Default = 2000).</pre>
*
* <pre> -C <number of epochs in convergence phase>
* Number of epochs to train through.
* (Value should be greater than or equal to 1000, Default = 1000).</pre>
*
* <pre> -H <height of lattice>
* The height of lattice.
* (Value should be > 0, Default = 1).</pre>
*
* <pre> -W <width of lattice>
* The width of lattice.
* (Value should be > 0, Default = 1).</pre>
*
* <pre> -I
* Normalizing the attributes will NOT be done.
* (Set this to not normalize the attributes).</pre>
*
* <pre> -S
* Statistics will NOT be calculated after training.
* (Set this to not calculate statistics).</pre>
*
<!-- options-end -->
*
* @author John Salatas (jsalatas at gmail.com)
* @version $Revision: 4 $
*/
public class SelfOrganizingMap extends AbstractClusterer
implements OptionHandler {
/** for serialization */
static final long serialVersionUID = -3028490959617832916L;
/** the distance function used. */
private EuclideanDistance m_euclideanDistance = new EuclideanDistance();
/** The width of lattice */
private int m_width;
/** The height of lattice */
private int m_height;
/** The number of epochs in convergence phase */
private int m_convergenceEpochs;
/** The number of epochs in ordering phase */
private int m_orderingEpochs;
/** The initial learning rate for the network */
private double m_learningRate;
/** The training instances. */
private Instances m_instances;
/** The weights for each unit in the lattice. */
private Instances m_clusters;
// It is used in EuclideanDistance.closestPoint
private int[] m_clusterList;
/** This flag states that the user wants the input values normalized. */
private boolean m_normalizeAttributes;
/** The maximum value for all the attributes. */
private double[] m_attributeMax;
/** The minimum value for all the attributes. */
private double[] m_attributeMin;
/** This flag states that the user wants to calculate statistics after training. */
private boolean m_calcStats;
/** holds the training instances to clusters assignments */
private Instances[] m_clusterInstances;
/** holds the cluster statistics */
private double[][][] m_clusterStats;
public static final int INPUT_COUNT=4;
public static final int SAMPLE_COUNT=100;
/**
* @return The width of lattice.
*/
public int getWidth() {
return m_width;
}
/**
* Sets the width of lattice.
* @param width The width of lattice.
*/
public void setWidth(int width) {
if (width > 0) {
this.m_width = width;
}
}
/**
* @return The height of lattice.
*/
public int getHeight() {
return m_height;
}
/**
* Sets the height of lattice.
* Must be greater than 0.
* @param height The height.
*/
public void setHeight(int height) {
if (height > 0) {
this.m_height = height;
}
}
/**
* @return The number of epochs in convergence phase.
*/
public int getConvergenceEpochs() {
return m_convergenceEpochs;
}
/**
* Set the number of epochs in convergence phase.
* Must be greater than or equal to 1000.
* @param n The number of epochs.
*/
public void setConvergenceEpochs(int n) {
if (n >= 1000) {
m_convergenceEpochs = n;
}
}
/**
* @return The number of epochs in ordering phase.
*/
public int getOrderingEpochs() {
return m_orderingEpochs;
}
/**
* Set the number of epochs in ordering phase.
* Must be greater than or equal to 2000.
* @param n The number of epochs.
*/
public void setOrderingEpochs(int n) {
if (n >= 2000) {
m_orderingEpochs = n;
}
}
/**
* @return The initial learning rate for the nodes.
*/
public double getLearningRate() {
return m_learningRate;
}
/**
* The initial learning rate can be set using this command.
* Must be greater than 0 and no more than 1.
* @param l The initial learning rate.
*/
public void setLearningRate(double l) {
if (l > 0.01 && l <= 1) {
m_learningRate = l;
}
}
/**
* @return The flag for normalizing attributes.
*/
public boolean getNormalizeAttributes() {
return m_normalizeAttributes;
}
/**
* @param a True if the attributes should be normalized (even nominal
* attributes will get normalized here) (range goes between -1 - 1).
*/
public void setNormalizeAttributes(boolean a) {
m_normalizeAttributes = a;
}
/**
* @return The flag for calculating statistics after training.
*/
public boolean getCalcStats() {
return m_calcStats;
}
/**
*
* @param c True if statistics should be calculated.
*/
public void setCalcStats(boolean c) {
this.m_calcStats = c;
}
/**
* @return a string to describe the height option.
*/
public String heightTipText() {
return "The height of lattice.";
}
/**
* @return a string to describe the caclulate statistics option.
*/
public String calcStatsTipText() {
return "This should calculate statistics for each cluster after training.";
}
/**
* @return a string to describe the width option.
*/
public String widthTipText() {
return "The width of lattice.";
}
/**
* @return a string to describe the learning rate option.
*/
public String learningRateTipText() {
return "The initial amount the weights are updated.";
}
/**
* @return a string to describe the number of epochs in convergence option.
*/
public String convergenceEpochsTipText() {
return "The number of epochs in convergence phase.";
}
/**
* @return a string to describe the number of epochs in ordering phase option.
*/
public String orderingEpochsTipText() {
return "The number of epochs in ordering phase.";
}
/**
* @return a string to describe the normalize attributes option.
*/
public String normalizeAttributesTipText() {
return "This will normalize the attributes.";
}
/**
* This will return a string describing the clusterer.
* @return The string.
*/
public String globalInfo() {
return "A Clusterer that implements Kohonen's Self Organizing Map\n"
+ "algorithm for unsupervised clustering.";
}
/**
* Returns the revision string.
*
* @return the revision
*/
public String getRevision() {
return RevisionUtils.extract("$Revision: 4 $");
}
/**
* The constructor.
*/
public SelfOrganizingMap() {
m_clusters = null;
m_width = 2;
m_height = 2;
m_convergenceEpochs = 1000;
m_orderingEpochs = 2000;
m_learningRate = 1.0;
m_normalizeAttributes = true;
m_calcStats = true;
}
/**
* Returns default capabilities of the classifier.
*
* @return the capabilities of this classifier
*/
public Capabilities getCapabilities() {
Capabilities result = super.getCapabilities();
result.disableAll();
result.enable(Capability.NO_CLASS);
// attributes
result.enable(Capability.NUMERIC_ATTRIBUTES);
result.enable(Capability.NOMINAL_ATTRIBUTES);
result.enable(Capability.MISSING_VALUES);
return result;
}
/**
* Parses a given list of options. <p/>
*
<!-- options-start -->
* Valid options are: <p/>
*
* <pre> -L <initial learning rate>
* The initial learning rate for the training algorithm.
* (Value should be greater than 0.01 and less or equal to 1, Default = 1).</pre>
*
* <pre> -O <number of epochs in ordering phase>
* Number of epochs in ordering phase.
* (Value should be greater than or equal to 2000, Default = 2000).</pre>
*
* <pre> -C <number of epochs in convergence phase>
* Number of epochs in convergence phase.
* (Value should be greater than or equal to 1000, Default = 1000).</pre>
*
* <pre> -I
* Normalizing the attributes will NOT be done.
* (Set this to not normalize the attributes).</pre>
*
* <pre> -H <height of lattice>
* The height of lattice.
* (Value should be > 0, Default = 2).</pre>
*
* <pre> -W <width of lattice>
* The width of lattice.
* (Value should be > 0, Default = 2).</pre>
*
* <pre> -S
* Statistics will NOT be calculated after training.
* (Set this to not calculate statistics).</pre>
*
<!-- options-end -->
*
* @param options the list of options as an array of strings
* @throws Exception if an option is not supported
*/
public void setOptions(String[] options) throws Exception {
//the defaults can be found here!!!!
String learningString = Utils.getOption('L', options);
if (learningString.length() != 0) {
setLearningRate((new Double(learningString)).doubleValue());
} else {
setLearningRate(1);
}
String orderingEpochsString = Utils.getOption('O', options);
if (orderingEpochsString.length() != 0) {
setOrderingEpochs(Integer.parseInt(orderingEpochsString));
} else {
setOrderingEpochs(2000);
}
String convergenceEpochsString = Utils.getOption('C', options);
if (convergenceEpochsString.length() != 0) {
setConvergenceEpochs(Integer.parseInt(convergenceEpochsString));
} else {
setConvergenceEpochs(1000);
}
String heightString = Utils.getOption('H', options);
if (heightString.length() != 0) {
setHeight(Integer.parseInt(heightString));
} else {
setHeight(2);
}
String widthString = Utils.getOption('W', options);
if (widthString.length() != 0) {
setWidth(Integer.parseInt(widthString));
} else {
setWidth(2);
}
if (Utils.getFlag('I', options)) {
setNormalizeAttributes(false);
} else {
setNormalizeAttributes(true);
}
if (Utils.getFlag('S', options)) {
setCalcStats(false);
} else {
setCalcStats(true);
}
Utils.checkForRemainingOptions(options);
}
/**
* Gets the current settings of NeuralNet.
*
* @return an array of strings suitable for passing to setOptions()
*/
public String[] getOptions() {
String[] options = new String[12];
int current = 0;
options[current++] = "-L";
options[current++] = "" + getLearningRate();
options[current++] = "-O";
options[current++] = "" + getOrderingEpochs();
options[current++] = "-C";
options[current++] = "" + getConvergenceEpochs();
options[current++] = "-H";
options[current++] = "" + getHeight();
options[current++] = "-W";
options[current++] = "" + getWidth();
if (!getNormalizeAttributes()) {
options[current++] = "-I";
}
if (!getCalcStats()) {
options[current++] = "-S";
}
while (current < options.length) {
options[current++] = "";
}
return options;
}
/**
* Classifies a given instance.
*
* @param i the instance to be assigned to a cluster
* @return the number of the assigned cluster as an interger
* if the class is enumerated, otherwise the predicted value
* @throws Exception if instance could not be classified
* successfully
*/
public int clusterInstance(Instance i) throws Exception {
if ((m_clusters == null) || (m_instances == null)) {
return 0;
}
Instance instance = new DenseInstance(i);
if (m_normalizeAttributes) {
instance = normalizeInstance(instance);
}
int a = m_euclideanDistance.closestPoint(instance, m_clusters, m_clusterList);
return a;
}
/**
* Returns an enumeration describing the available options.
*
* @return an enumeration of all the available options.
*/
public Enumeration listOptions() {
Vector result = new Vector();
result.addElement(new Option(
"\tLearning Rate for the training algorithm.\n"
+ "\t(default 1)",
"L", 1, "-L <num>"));
result.addElement(new Option(
"\tNumber of epochs in ordering phase.\n"
+ "\t(default 2000)",
"O", 1, "-O <num>"));
result.addElement(new Option(
"\tNumber of epochs in convergence phase.\n"
+ "\t(default 1000)",
"C", 1, "-C <num>"));
result.addElement(new Option(
"\tThe height of lattice.\n"
+ "\t(default 1)",
"H", 1, "-H <num>"));
result.addElement(new Option(
"\tThe width of lattice.\n"
+ "\t(default 1)",
"W", 1, "-W <num>"));
result.addElement(new Option(
"\tNormalizing the attributes will NOT be done.\n"
+ "\t(Set this to not normalize the attributes).",
"I", 0, "-I"));
return result.elements();
}
private String pad(String source, String padChar,
int length, boolean leftPad) {
StringBuffer temp = new StringBuffer();
if (leftPad) {
for (int i = 0; i < length; i++) {
temp.append(padChar);
}
temp.append(source);
} else {
temp.append(source);
for (int i = 0; i < length; i++) {
temp.append(padChar);
}
}
return temp.toString();
}
/**
* return a string describing this clusterer.
*
* @return a description of the clusterer as a string
*/
public String toString() {
StringBuilder sb = new StringBuilder();
sb.append("\nSelf Organized Map\n==================\n");
if ((m_clusters == null) || (m_instances == null)) {
sb.append("No clusterer built yet!\n");
sb.append("==================\n\n");
return sb.toString();
}
sb.append("\nNumber of clusters: " + (m_width * m_height) + "\n");
int maxWidth = 0;
int maxAttWidth = 0;
if (m_calcStats) {
// set up max widths
// attributes
for (int i = 0; i < m_clusters.numAttributes(); i++) {
Attribute a = m_clusters.attribute(i);
if (a.name().length() > maxAttWidth) {
maxAttWidth = m_clusters.attribute(i).name().length();
}
}
for (int i = 0; i < m_clusters.numInstances(); i++) {
for (int j = 0; j < m_clusters.numAttributes(); j++) {
// check mean and std. dev. against maxWidth
double mean = Math.log(Math.abs(m_clusterStats[j][i][2])) / Math.log(10.0);
double stdD = Math.log(Math.abs(m_clusterStats[j][i][3])) / Math.log(10.0);
double width = (mean > stdD)
? mean
: stdD;
if (width < 0) {
width = 1;
}
// decimal + # decimal places + 1
width += 6.0;
if ((int) width > maxWidth) {
maxWidth = (int) width;
}
}
}
if (maxAttWidth < "Attribute".length()) {
maxAttWidth = "Attribute".length();
}
maxAttWidth += 2;
sb.append("\n\n");
sb.append(pad("Cluster", " ",
(maxAttWidth + maxWidth + 1) - "Cluster".length(),
true));
sb.append("\n");
sb.append(pad("Attribute", " ", maxAttWidth - "Attribute".length(), false));
// cluster #'s
for (int i = 0; i < m_clusters.numInstances(); i++) {
String classL = "" + i;
sb.append(pad(classL, " ", maxWidth + 1 - classL.length(), true));
}
sb.append("\n");
sb.append(pad("", " ", maxAttWidth, true));
for (int i = 0; i < m_clusters.numInstances(); i++) {
String numInst = Utils.doubleToString(m_clusterInstances[i].numInstances(), maxWidth, 2).trim();
numInst = "(" + numInst + ")";
sb.append(pad(numInst, " ", maxWidth + 1 - numInst.length(), true));
}
sb.append("\n");
sb.append(pad("", "=", maxAttWidth
+ (maxWidth * m_clusters.numInstances())
+ m_clusters.numInstances() + 1, true));
sb.append("\n");
for (int i = 0; i < m_clusters.numAttributes(); i++) {
String attName = m_clusters.attribute(i).name();
sb.append(attName + "\n");
String valueL = " value";
sb.append(pad(valueL, " ", maxAttWidth + 1 - valueL.length(), false));
for (int j = 0; j < m_clusters.numInstances(); j++) {
// values
String value =
Utils.doubleToString(denormalizeInstance(m_clusters.get(j)).value(i), maxWidth, 4).trim();
sb.append(pad(value, " ", maxWidth + 1 - value.length(), true));
}
sb.append("\n");
String minL = " min";
sb.append(pad(minL, " ", maxAttWidth + 1 - minL.length(), false));
for (int j = 0; j < m_clusters.numInstances(); j++) {
// means
String min =
Utils.doubleToString(m_clusterStats[i][j][0], maxWidth, 4).trim();
sb.append(pad(min, " ", maxWidth + 1 - min.length(), true));
}
sb.append("\n");
String maxL = " max";
sb.append(pad(maxL, " ", maxAttWidth + 1 - maxL.length(), false));
for (int j = 0; j < m_clusters.numInstances(); j++) {
// means
String max =
Utils.doubleToString(m_clusterStats[i][j][1], maxWidth, 4).trim();
sb.append(pad(max, " ", maxWidth + 1 - max.length(), true));
}
sb.append("\n");
String meanL = " mean";
sb.append(pad(meanL, " ", maxAttWidth + 1 - meanL.length(), false));
for (int j = 0; j < m_clusters.numInstances(); j++) {
// means
String mean =
Utils.doubleToString(m_clusterStats[i][j][2], maxWidth, 4).trim();
sb.append(pad(mean, " ", maxWidth + 1 - mean.length(), true));
}
sb.append("\n");
// now do std deviations
String stdDevL = " std. dev.";
sb.append(pad(stdDevL, " ", maxAttWidth + 1 - stdDevL.length(), false));
for (int j = 0; j < m_clusters.numInstances(); j++) {
String stdDev =
Utils.doubleToString(m_clusterStats[i][j][3], maxWidth, 4).trim();
sb.append(pad(stdDev, " ", maxWidth + 1 - stdDev.length(), true));
}
sb.append("\n\n");
}
}
return sb.toString();
}
/**
* Generates a clusterer. Has to initialize all fields of the clusterer
* that are not being set via options.
*
* @param data set of instances serving as training data
* @throws Exception if the clusterer has not been
* generated successfully
*/
public void buildClusterer(Instances data) throws Exception {
// can clusterer handle the data?
getCapabilities().testWithFail(data);
// copy the original instances
m_instances = new Instances(data);
// normalize instances
m_instances = normalize(m_instances);
// init clusters
m_clusters = initClusters();
// init the pointList (used in EuclideanDistance.closestPoint)
m_clusterList = new int[m_clusters.numInstances()];
for (int i = 0; i < m_clusterList.length; i++) {
m_clusterList[i] = i;
}
// the actual learning rate
double learningRate;
// the actual effective neighborhood
double effectiveWidth;
// init euclidean distance
m_euclideanDistance.setDontNormalize(true);
m_euclideanDistance.setInstances(m_clusters);
// the winner neuron
int winningNeuron;
for (int epoch = 1; epoch <= m_convergenceEpochs + m_orderingEpochs; epoch++) {
learningRate = calcLearningRate(epoch);
effectiveWidth = calcEffectiveWidth(epoch);
for (int instance = 0; instance < m_instances.numInstances(); instance++) {
winningNeuron = m_euclideanDistance.closestPoint(m_instances.get(instance), m_clusters, m_clusterList);
// update the weights
for (int neuron = 0; neuron < m_clusters.numInstances(); neuron++) {
updateWeights(neuron, winningNeuron, instance, learningRate, effectiveWidth);
}
}
}
if (m_calcStats) {
calcStatistics();
}
}
/**
* This function calculates the clusterer's statistics
*/
private void calcStatistics() {
// init cluster statistics
m_clusterStats = new double[m_instances.numAttributes()][m_width * m_height][4];
// init clusters assignements
m_clusterInstances = new Instances[m_width * m_height];
// keep cluster's attributes
for (int i = 0; i < m_clusters.numInstances(); i++) {
m_clusterInstances[i] = new Instances(m_instances);
m_clusterInstances[i].clear();
try {
Instances clusters = getClusters();
for (int j = 0; j < clusters.numAttributes(); j++) {
m_clusterStats[j][i][0] = clusters.get(i).value(j);
}
} catch (Exception ex) {
}
}
// get instances in each class
for (int instance = 0; instance < m_instances.numInstances(); instance++) {
Instance inst = m_instances.get(instance);
int cluster = -1;
try {
cluster = clusterInstance(denormalizeInstance(inst));
} catch (Exception ex) {
}
if (cluster != -1) {
m_clusterInstances[cluster].add(denormalizeInstance(inst));
}
}
//calc min, max, mean, stdev for each cluster
for (int cluster = 0; cluster < m_clusters.numInstances(); cluster++) {
for (int attr = 0; attr < m_clusters.numAttributes(); attr++) {
int unknownValues = 0;
double min = Double.POSITIVE_INFINITY;
double max = Double.NEGATIVE_INFINITY;
double mean = 0;
double stdev = 0;
for (int instance = 0; instance < m_clusterInstances[cluster].numInstances(); instance++) {
Instance inst = m_clusterInstances[cluster].get(instance);
if (!Double.isNaN(inst.value(attr))) {
mean += inst.value(attr);
if (inst.value(attr) < min) {
min = inst.value(attr);
}
if (inst.value(attr) > max) {
max = inst.value(attr);
}
} else {
unknownValues++;
}
}
mean /= (m_clusterInstances[cluster].numInstances() - unknownValues);
for (int instance = 0; instance < m_clusterInstances[cluster].numInstances(); instance++) {
Instance inst = m_clusterInstances[cluster].get(instance);
if (!Double.isNaN(inst.value(attr))) {
stdev += (inst.value(attr) - mean) * (inst.value(attr) - mean);
}
}
stdev /= (m_clusterInstances[cluster].numInstances() - 1 - unknownValues);
stdev = Math.sqrt(stdev);
if (min == Double.POSITIVE_INFINITY) {
min = 0;
}
if (max == Double.NEGATIVE_INFINITY) {
max = 0;
}
if((m_clusterInstances[cluster].numInstances() - unknownValues)==0) {
min = Double.NaN;
max = Double.NaN;
stdev = Double.NaN;
}
m_clusterStats[attr][cluster][0] = min;
m_clusterStats[attr][cluster][1] = max;
m_clusterStats[attr][cluster][2] = mean;
m_clusterStats[attr][cluster][3] = stdev;
}
}
}
/**
* This function updates the weights of a cluster
* @param updateCluster the cluster to update
* @param winningCluster the winning cluster for the current training instance
* @param instance the index of current training instance
* @param learningRate the current learning rate
* @param effWidth the current effective width
*/
private void updateWeights(int updateCluster, int winningCluster, int instance, double learningRate, double effWidth) {
double diff;
int winnerWidth = winningCluster % m_width;
int winnerHeight = winningCluster / m_width;
int width = updateCluster % m_width;
int height = updateCluster / m_width;
double distance = Math.pow((winnerWidth - width), 2) + Math.pow((winnerHeight - height), 2);
double h = Math.exp(-distance / (2 * Math.pow(effWidth, 2)));
for (int j = 0; j < m_clusters.numAttributes(); j++) {
diff = learningRate * h * (m_instances.get(instance).value(j) - m_clusters.get(updateCluster).value(j));
if (!Double.isNaN(diff)) {
m_clusters.get(updateCluster).setValue(j, m_clusters.get(updateCluster).value(j) + diff);
}
}
}
/**
* This function calculates the effective width of the topological
* netowork for the current iteration
* @param iteration the current iteration.
* @return the effective width .
*/
private double calcEffectiveWidth(int iteration) {
double effectiveWidth;
if (iteration <= m_orderingEpochs) {
// Ordering phase
double latticeRadius = Math.sqrt(m_width * m_width + m_height * m_height) / 2;
effectiveWidth = latticeRadius * Math.exp(-iteration * Math.log(latticeRadius) / m_orderingEpochs);
} else {
// Convergence phase
effectiveWidth = 0.0001;
}
return effectiveWidth;
}
/**
* This function calculates the learning rate for the current iteration
* @param iteration the current iteration.
* @return the learning rate.
*/
private double calcLearningRate(int iteration) {
double learningRate;
if (iteration <= m_orderingEpochs) {
// Ordering phase
learningRate = m_learningRate * Math.exp(-(double) iteration * Math.log(100. * m_learningRate) / m_orderingEpochs);
} else {
// Convergence phase
learningRate = 0.01;
}
return learningRate;
}
/**
* This function performs the denormalization of the attributes of an instance.
*
* @param inst the instance.
* @return The modified instance. This needs to be done as it deep copies
* the instance which will need to be passed back out.
*/
protected Instance denormalizeInstance(Instance inst) {
inst = new DenseInstance(inst);
if (m_normalizeAttributes) {
for (int noa = 0; noa < inst.numAttributes(); noa++) {
inst.setValue(noa, (inst.value(noa) * (m_attributeMax[noa] - m_attributeMin[noa]) + (m_attributeMax[noa] + m_attributeMin[noa])) / 2);
}
}
return inst;
}
/**
* This function performs the normalization of the attributes of an instance.
*
* @param inst the instance.
* @return The modified instance. This needs to be done as it deep copies
* the instance which will need to be passed back out.
*/
protected Instance normalizeInstance(Instance inst) {
inst = new DenseInstance(inst);
double min;
double max;
for (int noa = 0; noa < m_instances.numAttributes(); noa++) {
if (inst.value(noa) > m_attributeMax[noa]) {
max = inst.value(noa);
} else {
max = m_attributeMax[noa];
}
if (inst.value(noa) < m_attributeMin[noa]) {
min = inst.value(noa);
} else {
min = m_attributeMin[noa];
}
if ((max - min) != 0) {
inst.setValue(noa, -1 + 2 * (inst.value(noa) - min) / (max - min));
} else {
inst.setValue(noa, inst.value(noa));
}
}
return inst;
}
/**
* This function performs the normalization of the attributes if applicable.
* (note that regardless of the options it will fill an array with the range
* and base, set to normalize all attributes and the class to be between -1
* and 1)
* @param inst the instances.
* @return The modified instances. This needs to be done. If the attributes
* are normalized then deep copies will be made of all the instances which
* will need to be passed back out.
*/
private Instances normalize(Instances inst) throws Exception {
if (inst != null) {
inst = new Instances(inst);
// x bounds
double min = Double.POSITIVE_INFINITY;
double max = Double.NEGATIVE_INFINITY;
double value;
m_attributeMax = new double[inst.numAttributes()];
m_attributeMin = new double[inst.numAttributes()];
for (int noa = 0; noa < inst.numAttributes(); noa++) {
min = Double.POSITIVE_INFINITY;
max = Double.NEGATIVE_INFINITY;
for (int i = 0; i < inst.numInstances(); i++) {
if (!inst.instance(i).isMissing(noa)) {
value = inst.instance(i).value(noa);
if (value < min) {
min = value;
}
if (value > max) {
max = value;
}
}
}
m_attributeMax[noa] = max;
m_attributeMin[noa] = min;
}
}
if (m_normalizeAttributes) {
for (int i = 0; i < inst.numInstances(); i++) {
inst.set(i, normalizeInstance(inst.instance(i)));
}
}
return inst;
}
/**
* This function initializes the clusters' weights.
*
* @return The initialized clusters
*/
protected Instances initClusters() {
Instances weights = new Instances(m_instances, m_width * m_height);
for (int i = 0; i < m_width * m_height; i++) {
double[] instValues = new double[m_instances.numAttributes()];
for (int j = 0; j < m_instances.numAttributes(); j++) {
if (m_normalizeAttributes) {
instValues[j] = 0;
} else {
instValues[j] = (m_attributeMax[j] + m_attributeMin[j]) / 2;
}
}
Instance inst = new DenseInstance(1, instValues);
weights.add(i, inst);
}
return weights;
}
/**
* Returns the number of clusters.
*
* @return the number of clusters generated for a training dataset.
* @exception Exception if number of clusters could not be returned
* successfully
*/
public int numberOfClusters() throws Exception {
return m_height * m_width;
}
/**
* This function returns the clusters if the clusterer is build
* or an exception if the clusterer is not build.
*
* The clusters are returned dernomalized even if normalizeAttributes
* option is set.
*
* @return The clusters
* @throws Exception
*/
public Instances getClusters() throws Exception {
if (m_clusters == null) {
throw new Exception("No clusterer built yet!");
}
Instances inst = new Instances(m_clusters);
if (m_normalizeAttributes) {
for (int i = 0; i < inst.numInstances(); i++) {
inst.set(i, denormalizeInstance(inst.instance(i)));
}
}
return inst;
}
/**
* This function returns the training statistics in a 3-dimension array
* as follows:<br/>
* First dimension: the attribute index<br/>
* Second dimension: the cluster index<br/>
* Third dimension: the static index (Valid values are 0: min, 1: max, 2: mean, 3: st. dev.)
*
* @return the statistics array
* @throws Exception
*/
public double[][][] getStatistics() throws Exception {
if (m_calcStats) {
if (m_clusterStats == null) {
throw new Exception("No clusterer built yet!");
}
} else {
throw new Exception("Statistics are not calculated");
}
return m_clusterStats;
}
/**
* This function returns the cluster assignment for each of the
* training instances. The array's index indicates the corresponding
* cluster.
*
* @return the cluster assignments array
* @throws Exception
*/
public Instances[] getClusterInstances() throws Exception {
if (m_calcStats) {
if (m_clusterInstances == null) {
throw new Exception("No clusterer built yet!");
}
} else {
throw new Exception("Statistics are not calculated");
}
return m_clusterInstances;
}
} |
Self Organizing Map
Répondre avec citation
Partager