1 | /* |
---|
2 | * This program is free software; you can redistribute it and/or modify |
---|
3 | * it under the terms of the GNU General Public License as published by |
---|
4 | * the Free Software Foundation; either version 2 of the License, or |
---|
5 | * (at your option) any later version. |
---|
6 | * |
---|
7 | * This program is distributed in the hope that it will be useful, |
---|
8 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
---|
9 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
---|
10 | * GNU General Public License for more details. |
---|
11 | * |
---|
12 | * You should have received a copy of the GNU General Public License |
---|
13 | * along with this program; if not, write to the Free Software |
---|
14 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
---|
15 | */ |
---|
16 | |
---|
17 | /* |
---|
18 | * ICSSearchAlgorithm.java |
---|
19 | * Copyright (C) 2004 University of Waikato, Hamilton, New Zealand |
---|
20 | * |
---|
21 | */ |
---|
22 | |
---|
23 | |
---|
24 | package weka.classifiers.bayes.net.search.ci; |
---|
25 | |
---|
26 | import weka.classifiers.bayes.BayesNet; |
---|
27 | import weka.classifiers.bayes.net.ParentSet; |
---|
28 | import weka.core.Instances; |
---|
29 | import weka.core.Option; |
---|
30 | import weka.core.RevisionHandler; |
---|
31 | import weka.core.RevisionUtils; |
---|
32 | import weka.core.Utils; |
---|
33 | |
---|
34 | import java.io.FileReader; |
---|
35 | import java.util.Enumeration; |
---|
36 | import java.util.Vector; |
---|
37 | |
---|
38 | /** |
---|
39 | <!-- globalinfo-start --> |
---|
40 | * This Bayes Network learning algorithm uses conditional independence tests to find a skeleton, finds V-nodes and applies a set of rules to find the directions of the remaining arrows. |
---|
41 | * <p/> |
---|
42 | <!-- globalinfo-end --> |
---|
43 | * |
---|
44 | <!-- options-start --> |
---|
45 | * Valid options are: <p/> |
---|
46 | * |
---|
47 | * <pre> -cardinality <num> |
---|
48 | * When determining whether an edge exists a search is performed |
---|
49 | * for a set Z that separates the nodes. MaxCardinality determines |
---|
50 | * the maximum size of the set Z. This greatly influences the |
---|
51 | * length of the search. (default 2)</pre> |
---|
52 | * |
---|
53 | * <pre> -mbc |
---|
54 | * Applies a Markov Blanket correction to the network structure, |
---|
55 | * after a network structure is learned. This ensures that all |
---|
56 | * nodes in the network are part of the Markov blanket of the |
---|
57 | * classifier node.</pre> |
---|
58 | * |
---|
59 | * <pre> -S [BAYES|MDL|ENTROPY|AIC|CROSS_CLASSIC|CROSS_BAYES] |
---|
60 | * Score type (BAYES, BDeu, MDL, ENTROPY and AIC)</pre> |
---|
61 | * |
---|
62 | <!-- options-end --> |
---|
63 | * |
---|
64 | * @author Remco Bouckaert |
---|
65 | * @version $Revision: 1.8 $ |
---|
66 | */ |
---|
67 | public class ICSSearchAlgorithm |
---|
68 | extends CISearchAlgorithm { |
---|
69 | |
---|
70 | /** for serialization */ |
---|
71 | static final long serialVersionUID = -2510985917284798576L; |
---|
72 | |
---|
73 | /** |
---|
74 | * returns the name of the attribute with the given index |
---|
75 | * |
---|
76 | * @param iAttribute the index of the attribute |
---|
77 | * @return the name of the attribute |
---|
78 | */ |
---|
79 | String name(int iAttribute) { |
---|
80 | return m_instances.attribute(iAttribute).name(); |
---|
81 | } |
---|
82 | |
---|
83 | /** |
---|
84 | * returns the number of attributes |
---|
85 | * |
---|
86 | * @return the number of attributes |
---|
87 | */ |
---|
88 | int maxn() { |
---|
89 | return m_instances.numAttributes(); |
---|
90 | } |
---|
91 | |
---|
92 | /** maximum size of separating set **/ |
---|
93 | private int m_nMaxCardinality = 2; |
---|
94 | |
---|
95 | /** |
---|
96 | * sets the cardinality |
---|
97 | * |
---|
98 | * @param nMaxCardinality the max cardinality |
---|
99 | */ |
---|
100 | public void setMaxCardinality(int nMaxCardinality) { |
---|
101 | m_nMaxCardinality = nMaxCardinality; |
---|
102 | } |
---|
103 | |
---|
104 | /** |
---|
105 | * returns the max cardinality |
---|
106 | * |
---|
107 | * @return the max cardinality |
---|
108 | */ |
---|
109 | public int getMaxCardinality() { |
---|
110 | return m_nMaxCardinality; |
---|
111 | } |
---|
112 | |
---|
113 | |
---|
114 | |
---|
115 | class SeparationSet |
---|
116 | implements RevisionHandler { |
---|
117 | |
---|
118 | public int [] m_set; |
---|
119 | |
---|
120 | /** |
---|
121 | * constructor |
---|
122 | */ |
---|
123 | public SeparationSet() { |
---|
124 | m_set= new int [getMaxCardinality() + 1]; |
---|
125 | } // c'tor |
---|
126 | |
---|
127 | public boolean contains(int nItem) { |
---|
128 | for (int iItem = 0; iItem < getMaxCardinality() && m_set[iItem] != -1; iItem++) { |
---|
129 | if (m_set[iItem] == nItem) { |
---|
130 | return true; |
---|
131 | } |
---|
132 | } |
---|
133 | return false; |
---|
134 | } // contains |
---|
135 | |
---|
136 | /** |
---|
137 | * Returns the revision string. |
---|
138 | * |
---|
139 | * @return the revision |
---|
140 | */ |
---|
141 | public String getRevision() { |
---|
142 | return RevisionUtils.extract("$Revision: 1.8 $"); |
---|
143 | } |
---|
144 | |
---|
145 | } // class sepset |
---|
146 | |
---|
147 | /** |
---|
148 | * Search for Bayes network structure using ICS algorithm |
---|
149 | * @param bayesNet datastructure to build network structure for |
---|
150 | * @param instances data set to learn from |
---|
151 | * @throws Exception if something goes wrong |
---|
152 | */ |
---|
153 | protected void search(BayesNet bayesNet, Instances instances) throws Exception { |
---|
154 | // init |
---|
155 | m_BayesNet = bayesNet; |
---|
156 | m_instances = instances; |
---|
157 | |
---|
158 | boolean edges[][] = new boolean [maxn() + 1][]; |
---|
159 | boolean [] [] arrows = new boolean [maxn() + 1][]; |
---|
160 | SeparationSet [] [] sepsets = new SeparationSet [maxn() + 1][]; |
---|
161 | for (int iNode = 0 ; iNode < maxn() + 1; iNode++) { |
---|
162 | edges[iNode] = new boolean[maxn()]; |
---|
163 | arrows[iNode] = new boolean[maxn()]; |
---|
164 | sepsets[iNode] = new SeparationSet[maxn()]; |
---|
165 | } |
---|
166 | |
---|
167 | calcDependencyGraph(edges, sepsets); |
---|
168 | calcVeeNodes(edges, arrows, sepsets); |
---|
169 | calcArcDirections(edges, arrows); |
---|
170 | |
---|
171 | // transfrom into BayesNet datastructure |
---|
172 | for (int iNode = 0; iNode < maxn(); iNode++) { |
---|
173 | // clear parent set of AttributeX |
---|
174 | ParentSet oParentSet = m_BayesNet.getParentSet(iNode); |
---|
175 | while (oParentSet.getNrOfParents() > 0) { |
---|
176 | oParentSet.deleteLastParent(m_instances); |
---|
177 | } |
---|
178 | for (int iParent = 0; iParent < maxn(); iParent++) { |
---|
179 | if (arrows[iParent][iNode]) { |
---|
180 | oParentSet.addParent(iParent, m_instances); |
---|
181 | } |
---|
182 | } |
---|
183 | } |
---|
184 | } // search |
---|
185 | |
---|
186 | |
---|
187 | /** CalcDependencyGraph determines the skeleton of the BayesNetwork by |
---|
188 | * starting with a complete graph and removing edges (a--b) if it can |
---|
189 | * find a set Z such that a and b conditionally independent given Z. |
---|
190 | * The set Z is found by trying all possible subsets of nodes adjacent |
---|
191 | * to a and b, first of size 0, then of size 1, etc. up to size |
---|
192 | * m_nMaxCardinality |
---|
193 | * @param edges boolean matrix representing the edges |
---|
194 | * @param sepsets set of separating sets |
---|
195 | */ |
---|
196 | void calcDependencyGraph(boolean[][] edges, SeparationSet[][] sepsets) { |
---|
197 | /*calc undirected graph a-b iff D(a,S,b) for all S)*/ |
---|
198 | SeparationSet oSepSet; |
---|
199 | |
---|
200 | for (int iNode1 = 0; iNode1 < maxn(); iNode1++) { |
---|
201 | /*start with complete graph*/ |
---|
202 | for (int iNode2 = 0; iNode2 < maxn(); iNode2++) { |
---|
203 | edges[iNode1][iNode2] = true; |
---|
204 | } |
---|
205 | } |
---|
206 | for (int iNode1 = 0; iNode1 < maxn(); iNode1++) { |
---|
207 | edges[iNode1][iNode1] = false; |
---|
208 | } |
---|
209 | |
---|
210 | for (int iCardinality = 0; iCardinality <= getMaxCardinality(); iCardinality++) { |
---|
211 | for (int iNode1 = 0; iNode1 <= maxn() - 2; iNode1++) { |
---|
212 | for (int iNode2 = iNode1 + 1; iNode2 < maxn(); iNode2++) { |
---|
213 | if (edges[iNode1][iNode2]) { |
---|
214 | oSepSet = existsSepSet(iNode1, iNode2, iCardinality, edges); |
---|
215 | if (oSepSet != null) { |
---|
216 | edges[iNode1][iNode2] = false; |
---|
217 | edges[iNode2][iNode1] = false; |
---|
218 | sepsets[iNode1][iNode2] = oSepSet; |
---|
219 | sepsets[iNode2][iNode1] = oSepSet; |
---|
220 | // report separating set |
---|
221 | System.err.print("I(" + name(iNode1) + ", {"); |
---|
222 | for (int iNode3 = 0; iNode3 < iCardinality; iNode3++) { |
---|
223 | System.err.print(name(oSepSet.m_set[iNode3]) + " "); |
---|
224 | } |
---|
225 | System.err.print("} ," + name(iNode2) + ")\n"); |
---|
226 | } |
---|
227 | } |
---|
228 | } |
---|
229 | } |
---|
230 | // report current state of dependency graph |
---|
231 | System.err.print(iCardinality + " "); |
---|
232 | for (int iNode1 = 0; iNode1 < maxn(); iNode1++) { |
---|
233 | System.err.print(name(iNode1) + " "); |
---|
234 | } |
---|
235 | System.err.print('\n'); |
---|
236 | for (int iNode1 = 0; iNode1 < maxn(); iNode1++) { |
---|
237 | for (int iNode2 = 0; iNode2 < maxn(); iNode2++) { |
---|
238 | if (edges[iNode1][iNode2]) |
---|
239 | System.err.print("X "); |
---|
240 | else |
---|
241 | System.err.print(". "); |
---|
242 | } |
---|
243 | System.err.print(name(iNode1) + " "); |
---|
244 | System.err.print('\n'); |
---|
245 | } |
---|
246 | } |
---|
247 | } /*CalcDependencyGraph*/ |
---|
248 | |
---|
249 | /** ExistsSepSet tests if a separating set Z of node a and b exists of given |
---|
250 | * cardiniality exists. |
---|
251 | * The set Z is found by trying all possible subsets of nodes adjacent |
---|
252 | * to both a and b of the requested cardinality. |
---|
253 | * @param iNode1 index of first node a |
---|
254 | * @param iNode2 index of second node b |
---|
255 | * @param nCardinality size of the separating set Z |
---|
256 | * @param edges |
---|
257 | * @return SeparationSet containing set that separates iNode1 and iNode2 or null if no such set exists |
---|
258 | */ |
---|
259 | SeparationSet existsSepSet(int iNode1, int iNode2, int nCardinality, boolean [] [] edges) |
---|
260 | { |
---|
261 | /*Test if a separating set of node d and e exists of cardiniality k*/ |
---|
262 | // int iNode1_, iNode2_; |
---|
263 | int iNode3, iZ; |
---|
264 | SeparationSet Z = new SeparationSet(); |
---|
265 | Z.m_set[nCardinality] = -1; |
---|
266 | |
---|
267 | // iNode1_ = iNode1; |
---|
268 | // iNode2_ = iNode2; |
---|
269 | |
---|
270 | // find first candidate separating set Z |
---|
271 | if (nCardinality > 0) { |
---|
272 | Z.m_set[0] = next(-1, iNode1, iNode2, edges); |
---|
273 | iNode3 = 1; |
---|
274 | while (iNode3 < nCardinality) { |
---|
275 | Z.m_set[iNode3] = next(Z.m_set[iNode3 - 1], iNode1, iNode2, edges); |
---|
276 | iNode3++; |
---|
277 | } |
---|
278 | } |
---|
279 | |
---|
280 | if (nCardinality > 0) { |
---|
281 | iZ = maxn() - Z.m_set[nCardinality - 1] - 1; |
---|
282 | } else { |
---|
283 | iZ = 0; |
---|
284 | } |
---|
285 | |
---|
286 | |
---|
287 | while (iZ >= 0) |
---|
288 | { |
---|
289 | //check if candidate separating set makes iNode2_ and iNode1_ independent |
---|
290 | if (isConditionalIndependent(iNode2, iNode1, Z.m_set, nCardinality)) { |
---|
291 | return Z; |
---|
292 | } |
---|
293 | // calc next candidate separating set |
---|
294 | if (nCardinality > 0) { |
---|
295 | Z.m_set[nCardinality - 1] = next(Z.m_set[nCardinality - 1], iNode1, iNode2, edges); |
---|
296 | } |
---|
297 | iZ = nCardinality - 1; |
---|
298 | while (iZ >= 0 && Z.m_set[iZ] >= maxn()) { |
---|
299 | iZ = nCardinality - 1; |
---|
300 | while (iZ >= 0 && Z.m_set[iZ] >= maxn()) { |
---|
301 | iZ--; |
---|
302 | } |
---|
303 | if (iZ < 0) { |
---|
304 | break; |
---|
305 | } |
---|
306 | Z.m_set[iZ] = next(Z.m_set[iZ], iNode1, iNode2, edges); |
---|
307 | for (iNode3 = iZ + 1; iNode3 < nCardinality; iNode3++) { |
---|
308 | Z.m_set[iNode3] = next(Z.m_set[iNode3 - 1], iNode1, iNode2, edges); |
---|
309 | } |
---|
310 | iZ = nCardinality - 1; |
---|
311 | } |
---|
312 | } |
---|
313 | |
---|
314 | return null; |
---|
315 | } /*ExistsSepSet*/ |
---|
316 | |
---|
317 | /** |
---|
318 | * determine index of node that makes next candidate separating set |
---|
319 | * adjacent to iNode1 and iNode2, but not iNode2 itself |
---|
320 | * @param x index of current node |
---|
321 | * @param iNode1 first node |
---|
322 | * @param iNode2 second node (must be larger than iNode1) |
---|
323 | * @param edges skeleton so far |
---|
324 | * @return int index of next node adjacent to iNode1 after x |
---|
325 | */ |
---|
326 | int next(int x, int iNode1, int iNode2, boolean [] [] edges) |
---|
327 | { |
---|
328 | x++; |
---|
329 | while (x < maxn() && (!edges[iNode1][x] || !edges[iNode2][x] ||x == iNode2)) { |
---|
330 | x++; |
---|
331 | } |
---|
332 | return x; |
---|
333 | } /*next*/ |
---|
334 | |
---|
335 | |
---|
336 | /** CalcVeeNodes tries to find V-nodes, i.e. nodes a,b,c such that |
---|
337 | * a->c<-b and a-/-b. These nodes are identified by finding nodes |
---|
338 | * a,b,c in the skeleton such that a--c, c--b and a-/-b and furthermore |
---|
339 | * c is not in the set Z that separates a and b |
---|
340 | * @param edges skeleton |
---|
341 | * @param arrows resulting partially directed skeleton after all V-nodes |
---|
342 | * have been identified |
---|
343 | * @param sepsets separating sets |
---|
344 | */ |
---|
345 | void calcVeeNodes( |
---|
346 | boolean[][] edges, |
---|
347 | boolean[][] arrows, |
---|
348 | SeparationSet[][] sepsets) { |
---|
349 | |
---|
350 | // start with complete empty graph |
---|
351 | for (int iNode1 = 0; iNode1 < maxn(); iNode1++) { |
---|
352 | for (int iNode2 = 0; iNode2 < maxn(); iNode2++) { |
---|
353 | arrows[iNode1][iNode2] = false; |
---|
354 | } |
---|
355 | } |
---|
356 | |
---|
357 | for (int iNode1 = 0; iNode1 < maxn() - 1; iNode1++) { |
---|
358 | for (int iNode2 = iNode1 + 1; iNode2 < maxn(); iNode2++) { |
---|
359 | if (!edges[iNode1][iNode2]) { /*i nonadj j*/ |
---|
360 | for (int iNode3 = 0; iNode3 < maxn(); iNode3++) { |
---|
361 | if ((iNode3 != iNode1 |
---|
362 | && iNode3 != iNode2 |
---|
363 | && edges[iNode1][iNode3] |
---|
364 | && edges[iNode2][iNode3]) |
---|
365 | & (!sepsets[iNode1][iNode2].contains(iNode3))) { |
---|
366 | arrows[iNode1][iNode3] = true; /*add arc i->k*/ |
---|
367 | arrows[iNode2][iNode3] = true; /*add arc j->k*/ |
---|
368 | } |
---|
369 | } |
---|
370 | } |
---|
371 | } |
---|
372 | } |
---|
373 | } // CalcVeeNodes |
---|
374 | |
---|
375 | |
---|
376 | /** CalcArcDirections assigns directions to edges that remain after V-nodes have |
---|
377 | * been identified. The arcs are directed using the following rules: |
---|
378 | Rule 1: i->j--k & i-/-k => j->k |
---|
379 | Rule 2: i->j->k & i--k => i->k |
---|
380 | Rule 3 m |
---|
381 | /|\ |
---|
382 | i | k => m->j |
---|
383 | i->j<-k \|/ |
---|
384 | j |
---|
385 | |
---|
386 | Rule 4 m |
---|
387 | / \ |
---|
388 | i---k => i->m & k->m |
---|
389 | i->j \ / |
---|
390 | j |
---|
391 | Rule 5: if no edges are directed then take a random one (first we can find) |
---|
392 | * @param edges skeleton |
---|
393 | * @param arrows resulting fully directed DAG |
---|
394 | */ |
---|
395 | void calcArcDirections(boolean[][] edges, boolean[][] arrows) { |
---|
396 | /*give direction to remaining arcs*/ |
---|
397 | int i, j, k, m; |
---|
398 | boolean bFound; |
---|
399 | |
---|
400 | do { |
---|
401 | bFound = false; |
---|
402 | |
---|
403 | /*Rule 1: i->j--k & i-/-k => j->k*/ |
---|
404 | |
---|
405 | for (i = 0; i < maxn(); i++) { |
---|
406 | for (j = 0; j < maxn(); j++) { |
---|
407 | if (i != j && arrows[i][j]) { |
---|
408 | for (k = 0; k < maxn(); k++) { |
---|
409 | if (i != k |
---|
410 | && j != k |
---|
411 | && edges[j][k] |
---|
412 | && !edges[i][k] |
---|
413 | && !arrows[j][k] |
---|
414 | && !arrows[k][j]) { |
---|
415 | arrows[j][k] = true; |
---|
416 | bFound = true; |
---|
417 | } |
---|
418 | } |
---|
419 | } |
---|
420 | } |
---|
421 | } |
---|
422 | |
---|
423 | /*Rule 2: i->j->k & i--k => i->k*/ |
---|
424 | |
---|
425 | for (i = 0; i < maxn(); i++) { |
---|
426 | for (j = 0; j < maxn(); j++) { |
---|
427 | if (i != j && arrows[i][j]) { |
---|
428 | for (k = 0; k < maxn(); k++) { |
---|
429 | if (i != k |
---|
430 | && j != k |
---|
431 | && edges[i][k] |
---|
432 | && arrows[j][k] |
---|
433 | && !arrows[i][k] |
---|
434 | && !arrows[k][i]) { |
---|
435 | arrows[i][k] = true; |
---|
436 | bFound = true; |
---|
437 | } |
---|
438 | } |
---|
439 | } |
---|
440 | } |
---|
441 | } |
---|
442 | |
---|
443 | /* Rule 3 m |
---|
444 | /|\ |
---|
445 | i | k => m->j |
---|
446 | i->j<-k \|/ |
---|
447 | j |
---|
448 | */ |
---|
449 | for (i = 0; i < maxn(); i++) { |
---|
450 | for (j = 0; j < maxn(); j++) { |
---|
451 | if (i != j && arrows[i][j]) { |
---|
452 | for (k = 0; k < maxn(); k++) { |
---|
453 | if (k != i |
---|
454 | && k != j |
---|
455 | && arrows[k][j] |
---|
456 | && !edges[k][i]) { |
---|
457 | for (m = 0; m < maxn(); m++) { |
---|
458 | if (m != i |
---|
459 | && m != j |
---|
460 | && m != k |
---|
461 | && edges[m][i] |
---|
462 | && !arrows[m][i] |
---|
463 | && !arrows[i][m] |
---|
464 | && edges[m][j] |
---|
465 | && !arrows[m][j] |
---|
466 | && !arrows[j][m] |
---|
467 | && edges[m][k] |
---|
468 | && !arrows[m][k] |
---|
469 | && !arrows[k][m]) { |
---|
470 | arrows[m][j] = true; |
---|
471 | bFound = true; |
---|
472 | } |
---|
473 | } |
---|
474 | } |
---|
475 | } |
---|
476 | } |
---|
477 | } |
---|
478 | } |
---|
479 | |
---|
480 | /* Rule 4 m |
---|
481 | / \ |
---|
482 | i---k => i->m & k->m |
---|
483 | i->j \ / |
---|
484 | j |
---|
485 | */ |
---|
486 | for (i = 0; i < maxn(); i++) { |
---|
487 | for (j = 0; j < maxn(); j++) { |
---|
488 | if (i != j && arrows[j][i]) { |
---|
489 | for (k = 0; k < maxn(); k++) { |
---|
490 | if (k != i |
---|
491 | && k != j |
---|
492 | && edges[k][j] |
---|
493 | && !arrows[k][j] |
---|
494 | && !arrows[j][k] |
---|
495 | && edges[k][i] |
---|
496 | && !arrows[k][i] |
---|
497 | && !arrows[i][k]) { |
---|
498 | for (m = 0; m < maxn(); m++) { |
---|
499 | if (m != i |
---|
500 | && m != j |
---|
501 | && m != k |
---|
502 | && edges[m][i] |
---|
503 | && !arrows[m][i] |
---|
504 | && !arrows[i][m] |
---|
505 | && edges[m][k] |
---|
506 | && !arrows[m][k] |
---|
507 | && !arrows[k][m]) { |
---|
508 | arrows[i][m] = true; |
---|
509 | arrows[k][m] = true; |
---|
510 | bFound = true; |
---|
511 | } |
---|
512 | } |
---|
513 | } |
---|
514 | } |
---|
515 | } |
---|
516 | } |
---|
517 | } |
---|
518 | |
---|
519 | /*Rule 5: if no edges are directed then take a random one (first we can find)*/ |
---|
520 | |
---|
521 | if (!bFound) { |
---|
522 | i = 0; |
---|
523 | while (!bFound && i < maxn()) { |
---|
524 | j = 0; |
---|
525 | while (!bFound && j < maxn()) { |
---|
526 | if (edges[i][j] |
---|
527 | && !arrows[i][j] |
---|
528 | && !arrows[j][i]) { |
---|
529 | arrows[i][j] = true; |
---|
530 | bFound = true; |
---|
531 | } |
---|
532 | j++; |
---|
533 | } |
---|
534 | i++; |
---|
535 | } |
---|
536 | } |
---|
537 | |
---|
538 | } |
---|
539 | while (bFound); |
---|
540 | |
---|
541 | } // CalcArcDirections |
---|
542 | |
---|
543 | /** |
---|
544 | * Returns an enumeration describing the available options. |
---|
545 | * |
---|
546 | * @return an enumeration of all the available options. |
---|
547 | */ |
---|
548 | public Enumeration listOptions() { |
---|
549 | Vector result = new Vector(); |
---|
550 | |
---|
551 | result.addElement(new Option( |
---|
552 | "\tWhen determining whether an edge exists a search is performed \n" |
---|
553 | + "\tfor a set Z that separates the nodes. MaxCardinality determines \n" |
---|
554 | + "\tthe maximum size of the set Z. This greatly influences the \n" |
---|
555 | + "\tlength of the search. (default 2)", |
---|
556 | "cardinality", 1, "-cardinality <num>")); |
---|
557 | |
---|
558 | Enumeration en = super.listOptions(); |
---|
559 | while (en.hasMoreElements()) |
---|
560 | result.addElement(en.nextElement()); |
---|
561 | |
---|
562 | return result.elements(); |
---|
563 | } // listOption |
---|
564 | |
---|
565 | /** |
---|
566 | * Parses a given list of options. <p/> |
---|
567 | * |
---|
568 | <!-- options-start --> |
---|
569 | * Valid options are: <p/> |
---|
570 | * |
---|
571 | * <pre> -cardinality <num> |
---|
572 | * When determining whether an edge exists a search is performed |
---|
573 | * for a set Z that separates the nodes. MaxCardinality determines |
---|
574 | * the maximum size of the set Z. This greatly influences the |
---|
575 | * length of the search. (default 2)</pre> |
---|
576 | * |
---|
577 | * <pre> -mbc |
---|
578 | * Applies a Markov Blanket correction to the network structure, |
---|
579 | * after a network structure is learned. This ensures that all |
---|
580 | * nodes in the network are part of the Markov blanket of the |
---|
581 | * classifier node.</pre> |
---|
582 | * |
---|
583 | * <pre> -S [BAYES|MDL|ENTROPY|AIC|CROSS_CLASSIC|CROSS_BAYES] |
---|
584 | * Score type (BAYES, BDeu, MDL, ENTROPY and AIC)</pre> |
---|
585 | * |
---|
586 | <!-- options-end --> |
---|
587 | * |
---|
588 | * @param options the list of options as an array of strings |
---|
589 | * @throws Exception if an option is not supported |
---|
590 | */ |
---|
591 | public void setOptions(String[] options) throws Exception { |
---|
592 | String tmpStr; |
---|
593 | |
---|
594 | tmpStr = Utils.getOption("cardinality", options); |
---|
595 | if (tmpStr.length() != 0) |
---|
596 | setMaxCardinality(Integer.parseInt(tmpStr)); |
---|
597 | else |
---|
598 | setMaxCardinality(2); |
---|
599 | |
---|
600 | super.setOptions(options); |
---|
601 | } // setOptions |
---|
602 | |
---|
603 | /** |
---|
604 | * Gets the current settings of the Classifier. |
---|
605 | * |
---|
606 | * @return an array of strings suitable for passing to setOptions |
---|
607 | */ |
---|
608 | public String[] getOptions() { |
---|
609 | Vector result; |
---|
610 | String[] options; |
---|
611 | int i; |
---|
612 | |
---|
613 | result = new Vector(); |
---|
614 | options = super.getOptions(); |
---|
615 | for (i = 0; i < options.length; i++) |
---|
616 | result.add(options[i]); |
---|
617 | |
---|
618 | result.add("-cardinality"); |
---|
619 | result.add("" + getMaxCardinality()); |
---|
620 | |
---|
621 | return (String[]) result.toArray(new String[result.size()]); |
---|
622 | } // getOptions |
---|
623 | |
---|
624 | |
---|
625 | /** |
---|
626 | * @return a string to describe the MaxCardinality option. |
---|
627 | */ |
---|
628 | public String maxCardinalityTipText() { |
---|
629 | return "When determining whether an edge exists a search is performed for a set Z "+ |
---|
630 | "that separates the nodes. MaxCardinality determines the maximum size of the set Z. " + |
---|
631 | "This greatly influences the length of the search. Default value is 2."; |
---|
632 | } // maxCardinalityTipText |
---|
633 | |
---|
634 | /** |
---|
635 | * This will return a string describing the search algorithm. |
---|
636 | * @return The string. |
---|
637 | */ |
---|
638 | public String globalInfo() { |
---|
639 | return "This Bayes Network learning algorithm uses conditional independence tests " + |
---|
640 | "to find a skeleton, finds V-nodes and applies a set of rules to find the directions " + |
---|
641 | "of the remaining arrows."; |
---|
642 | } |
---|
643 | |
---|
644 | /** |
---|
645 | * Returns the revision string. |
---|
646 | * |
---|
647 | * @return the revision |
---|
648 | */ |
---|
649 | public String getRevision() { |
---|
650 | return RevisionUtils.extract("$Revision: 1.8 $"); |
---|
651 | } |
---|
652 | |
---|
653 | /** |
---|
654 | * for testing the class |
---|
655 | * |
---|
656 | * @param argv the commandline parameters |
---|
657 | */ |
---|
658 | static public void main(String [] argv) { |
---|
659 | try { |
---|
660 | BayesNet b = new BayesNet(); |
---|
661 | b.setSearchAlgorithm( new ICSSearchAlgorithm()); |
---|
662 | Instances instances = new Instances(new FileReader("C:\\eclipse\\workspace\\weka\\data\\contact-lenses.arff")); |
---|
663 | instances.setClassIndex(instances.numAttributes() - 1); |
---|
664 | b.buildClassifier(instances); |
---|
665 | System.out.println(b.toString()); |
---|
666 | } catch (Exception e) { |
---|
667 | e.printStackTrace(); |
---|
668 | } |
---|
669 | } // main |
---|
670 | |
---|
671 | } // class ICSSearchAlgorithm |
---|