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 | * AddClassification.java |
---|
19 | * Copyright (C) 2006 University of Waikato, Hamilton, New Zealand |
---|
20 | */ |
---|
21 | |
---|
22 | package weka.filters.supervised.attribute; |
---|
23 | |
---|
24 | import weka.classifiers.Classifier; |
---|
25 | import weka.classifiers.AbstractClassifier; |
---|
26 | import weka.core.Attribute; |
---|
27 | import weka.core.Capabilities; |
---|
28 | import weka.core.FastVector; |
---|
29 | import weka.core.Instance; |
---|
30 | import weka.core.DenseInstance; |
---|
31 | import weka.core.Instances; |
---|
32 | import weka.core.Option; |
---|
33 | import weka.core.OptionHandler; |
---|
34 | import weka.core.RevisionUtils; |
---|
35 | import weka.core.SparseInstance; |
---|
36 | import weka.core.Utils; |
---|
37 | import weka.core.WekaException; |
---|
38 | import weka.filters.SimpleBatchFilter; |
---|
39 | |
---|
40 | import java.io.File; |
---|
41 | import java.io.FileInputStream; |
---|
42 | import java.io.FileNotFoundException; |
---|
43 | import java.io.ObjectInputStream; |
---|
44 | import java.util.Enumeration; |
---|
45 | import java.util.Vector; |
---|
46 | |
---|
47 | /** |
---|
48 | <!-- globalinfo-start --> |
---|
49 | * A filter for adding the classification, the class distribution and an error flag to a dataset with a classifier. The classifier is either trained on the data itself or provided as serialized model. |
---|
50 | * <p/> |
---|
51 | <!-- globalinfo-end --> |
---|
52 | * |
---|
53 | <!-- options-start --> |
---|
54 | * Valid options are: <p/> |
---|
55 | * |
---|
56 | * <pre> -D |
---|
57 | * Turns on output of debugging information.</pre> |
---|
58 | * |
---|
59 | * <pre> -W <classifier specification> |
---|
60 | * Full class name of classifier to use, followed |
---|
61 | * by scheme options. eg: |
---|
62 | * "weka.classifiers.bayes.NaiveBayes -D" |
---|
63 | * (default: weka.classifiers.rules.ZeroR)</pre> |
---|
64 | * |
---|
65 | * <pre> -serialized <file> |
---|
66 | * Instead of training a classifier on the data, one can also provide |
---|
67 | * a serialized model and use that for tagging the data.</pre> |
---|
68 | * |
---|
69 | * <pre> -classification |
---|
70 | * Adds an attribute with the actual classification. |
---|
71 | * (default: off)</pre> |
---|
72 | * |
---|
73 | * <pre> -remove-old-class |
---|
74 | * Removes the old class attribute. |
---|
75 | * (default: off)</pre> |
---|
76 | * |
---|
77 | * <pre> -distribution |
---|
78 | * Adds attributes with the distribution for all classes |
---|
79 | * (for numeric classes this will be identical to the attribute |
---|
80 | * output with '-classification'). |
---|
81 | * (default: off)</pre> |
---|
82 | * |
---|
83 | * <pre> -error |
---|
84 | * Adds an attribute indicating whether the classifier output |
---|
85 | * a wrong classification (for numeric classes this is the numeric |
---|
86 | * difference). |
---|
87 | * (default: off)</pre> |
---|
88 | * |
---|
89 | <!-- options-end --> |
---|
90 | * |
---|
91 | * @author fracpete (fracpete at waikato dot ac dot nz) |
---|
92 | * @version $Revision: 5987 $ |
---|
93 | */ |
---|
94 | public class AddClassification |
---|
95 | extends SimpleBatchFilter { |
---|
96 | |
---|
97 | /** for serialization */ |
---|
98 | private static final long serialVersionUID = -1931467132568441909L; |
---|
99 | |
---|
100 | /** The classifier template used to do the classification */ |
---|
101 | protected Classifier m_Classifier = new weka.classifiers.rules.ZeroR(); |
---|
102 | |
---|
103 | /** The file from which to load a serialized classifier */ |
---|
104 | protected File m_SerializedClassifierFile = new File(System.getProperty("user.dir")); |
---|
105 | |
---|
106 | /** The actual classifier used to do the classification */ |
---|
107 | protected Classifier m_ActualClassifier = null; |
---|
108 | |
---|
109 | /** whether to output the classification */ |
---|
110 | protected boolean m_OutputClassification = false; |
---|
111 | |
---|
112 | /** whether to remove the old class attribute */ |
---|
113 | protected boolean m_RemoveOldClass = false; |
---|
114 | |
---|
115 | /** whether to output the class distribution */ |
---|
116 | protected boolean m_OutputDistribution = false; |
---|
117 | |
---|
118 | /** whether to output the error flag */ |
---|
119 | protected boolean m_OutputErrorFlag = false; |
---|
120 | |
---|
121 | /** |
---|
122 | * Returns a string describing this filter |
---|
123 | * |
---|
124 | * @return a description of the filter suitable for |
---|
125 | * displaying in the explorer/experimenter gui |
---|
126 | */ |
---|
127 | public String globalInfo() { |
---|
128 | return |
---|
129 | "A filter for adding the classification, the class distribution and " |
---|
130 | + "an error flag to a dataset with a classifier. The classifier is " |
---|
131 | + "either trained on the data itself or provided as serialized model."; |
---|
132 | } |
---|
133 | |
---|
134 | /** |
---|
135 | * Returns an enumeration describing the available options. |
---|
136 | * |
---|
137 | * @return an enumeration of all the available options. |
---|
138 | */ |
---|
139 | public Enumeration listOptions() { |
---|
140 | Vector result; |
---|
141 | Enumeration en; |
---|
142 | |
---|
143 | result = new Vector(); |
---|
144 | |
---|
145 | en = super.listOptions(); |
---|
146 | while (en.hasMoreElements()) |
---|
147 | result.addElement(en.nextElement()); |
---|
148 | |
---|
149 | result.addElement(new Option( |
---|
150 | "\tFull class name of classifier to use, followed\n" |
---|
151 | + "\tby scheme options. eg:\n" |
---|
152 | + "\t\t\"weka.classifiers.bayes.NaiveBayes -D\"\n" |
---|
153 | + "\t(default: weka.classifiers.rules.ZeroR)", |
---|
154 | "W", 1, "-W <classifier specification>")); |
---|
155 | |
---|
156 | result.addElement(new Option( |
---|
157 | "\tInstead of training a classifier on the data, one can also provide\n" |
---|
158 | + "\ta serialized model and use that for tagging the data.", |
---|
159 | "serialized", 1, "-serialized <file>")); |
---|
160 | |
---|
161 | result.addElement(new Option( |
---|
162 | "\tAdds an attribute with the actual classification.\n" |
---|
163 | + "\t(default: off)", |
---|
164 | "classification", 0, "-classification")); |
---|
165 | |
---|
166 | result.addElement(new Option( |
---|
167 | "\tRemoves the old class attribute.\n" |
---|
168 | + "\t(default: off)", |
---|
169 | "remove-old-class", 0, "-remove-old-class")); |
---|
170 | |
---|
171 | result.addElement(new Option( |
---|
172 | "\tAdds attributes with the distribution for all classes \n" |
---|
173 | + "\t(for numeric classes this will be identical to the attribute \n" |
---|
174 | + "\toutput with '-classification').\n" |
---|
175 | + "\t(default: off)", |
---|
176 | "distribution", 0, "-distribution")); |
---|
177 | |
---|
178 | result.addElement(new Option( |
---|
179 | "\tAdds an attribute indicating whether the classifier output \n" |
---|
180 | + "\ta wrong classification (for numeric classes this is the numeric \n" |
---|
181 | + "\tdifference).\n" |
---|
182 | + "\t(default: off)", |
---|
183 | "error", 0, "-error")); |
---|
184 | |
---|
185 | return result.elements(); |
---|
186 | } |
---|
187 | |
---|
188 | /** |
---|
189 | * Parses the options for this object. <p/> |
---|
190 | * |
---|
191 | <!-- options-start --> |
---|
192 | * Valid options are: <p/> |
---|
193 | * |
---|
194 | * <pre> -D |
---|
195 | * Turns on output of debugging information.</pre> |
---|
196 | * |
---|
197 | * <pre> -W <classifier specification> |
---|
198 | * Full class name of classifier to use, followed |
---|
199 | * by scheme options. eg: |
---|
200 | * "weka.classifiers.bayes.NaiveBayes -D" |
---|
201 | * (default: weka.classifiers.rules.ZeroR)</pre> |
---|
202 | * |
---|
203 | * <pre> -serialized <file> |
---|
204 | * Instead of training a classifier on the data, one can also provide |
---|
205 | * a serialized model and use that for tagging the data.</pre> |
---|
206 | * |
---|
207 | * <pre> -classification |
---|
208 | * Adds an attribute with the actual classification. |
---|
209 | * (default: off)</pre> |
---|
210 | * |
---|
211 | * <pre> -remove-old-class |
---|
212 | * Removes the old class attribute. |
---|
213 | * (default: off)</pre> |
---|
214 | * |
---|
215 | * <pre> -distribution |
---|
216 | * Adds attributes with the distribution for all classes |
---|
217 | * (for numeric classes this will be identical to the attribute |
---|
218 | * output with '-classification'). |
---|
219 | * (default: off)</pre> |
---|
220 | * |
---|
221 | * <pre> -error |
---|
222 | * Adds an attribute indicating whether the classifier output |
---|
223 | * a wrong classification (for numeric classes this is the numeric |
---|
224 | * difference). |
---|
225 | * (default: off)</pre> |
---|
226 | * |
---|
227 | <!-- options-end --> |
---|
228 | * |
---|
229 | * @param options the options to use |
---|
230 | * @throws Exception if setting of options fails |
---|
231 | */ |
---|
232 | public void setOptions(String[] options) throws Exception { |
---|
233 | String tmpStr; |
---|
234 | String[] tmpOptions; |
---|
235 | File file; |
---|
236 | boolean serializedModel; |
---|
237 | |
---|
238 | setOutputClassification(Utils.getFlag("classification", options)); |
---|
239 | |
---|
240 | setRemoveOldClass(Utils.getFlag("remove-old-class", options)); |
---|
241 | |
---|
242 | setOutputDistribution(Utils.getFlag("distribution", options)); |
---|
243 | |
---|
244 | setOutputErrorFlag(Utils.getFlag("error", options)); |
---|
245 | |
---|
246 | serializedModel = false; |
---|
247 | tmpStr = Utils.getOption("serialized", options); |
---|
248 | if (tmpStr.length() != 0) { |
---|
249 | file = new File(tmpStr); |
---|
250 | if (!file.exists()) |
---|
251 | throw new FileNotFoundException( |
---|
252 | "File '" + file.getAbsolutePath() + "' not found!"); |
---|
253 | if (file.isDirectory()) |
---|
254 | throw new FileNotFoundException( |
---|
255 | "'" + file.getAbsolutePath() + "' points to a directory not a file!"); |
---|
256 | setSerializedClassifierFile(file); |
---|
257 | serializedModel = true; |
---|
258 | } |
---|
259 | else { |
---|
260 | setSerializedClassifierFile(null); |
---|
261 | } |
---|
262 | |
---|
263 | if (!serializedModel) { |
---|
264 | tmpStr = Utils.getOption('W', options); |
---|
265 | if (tmpStr.length() == 0) |
---|
266 | tmpStr = weka.classifiers.rules.ZeroR.class.getName(); |
---|
267 | tmpOptions = Utils.splitOptions(tmpStr); |
---|
268 | if (tmpOptions.length == 0) |
---|
269 | throw new Exception("Invalid classifier specification string"); |
---|
270 | tmpStr = tmpOptions[0]; |
---|
271 | tmpOptions[0] = ""; |
---|
272 | setClassifier(AbstractClassifier.forName(tmpStr, tmpOptions)); |
---|
273 | } |
---|
274 | |
---|
275 | super.setOptions(options); |
---|
276 | } |
---|
277 | |
---|
278 | /** |
---|
279 | * Gets the current settings of the classifier. |
---|
280 | * |
---|
281 | * @return an array of strings suitable for passing to setOptions |
---|
282 | */ |
---|
283 | public String[] getOptions() { |
---|
284 | int i; |
---|
285 | Vector result; |
---|
286 | String[] options; |
---|
287 | File file; |
---|
288 | |
---|
289 | result = new Vector(); |
---|
290 | |
---|
291 | options = super.getOptions(); |
---|
292 | for (i = 0; i < options.length; i++) |
---|
293 | result.add(options[i]); |
---|
294 | |
---|
295 | if (getOutputClassification()) |
---|
296 | result.add("-classification"); |
---|
297 | |
---|
298 | if (getRemoveOldClass()) |
---|
299 | result.add("-remove-old-class"); |
---|
300 | |
---|
301 | if (getOutputDistribution()) |
---|
302 | result.add("-distribution"); |
---|
303 | |
---|
304 | if (getOutputErrorFlag()) |
---|
305 | result.add("-error"); |
---|
306 | |
---|
307 | file = getSerializedClassifierFile(); |
---|
308 | if ((file != null) && (!file.isDirectory())) { |
---|
309 | result.add("-serialized"); |
---|
310 | result.add(file.getAbsolutePath()); |
---|
311 | } |
---|
312 | else { |
---|
313 | result.add("-W"); |
---|
314 | result.add(getClassifierSpec()); |
---|
315 | } |
---|
316 | |
---|
317 | return (String[]) result.toArray(new String[result.size()]); |
---|
318 | } |
---|
319 | |
---|
320 | /** |
---|
321 | * Returns the Capabilities of this filter. |
---|
322 | * |
---|
323 | * @return the capabilities of this object |
---|
324 | * @see Capabilities |
---|
325 | */ |
---|
326 | public Capabilities getCapabilities() { |
---|
327 | Capabilities result; |
---|
328 | |
---|
329 | if (getClassifier() == null) { |
---|
330 | result = super.getCapabilities(); |
---|
331 | result.disableAll(); |
---|
332 | } else { |
---|
333 | result = getClassifier().getCapabilities(); |
---|
334 | } |
---|
335 | |
---|
336 | result.setMinimumNumberInstances(0); |
---|
337 | |
---|
338 | return result; |
---|
339 | } |
---|
340 | |
---|
341 | /** |
---|
342 | * Returns the tip text for this property |
---|
343 | * |
---|
344 | * @return tip text for this property suitable for |
---|
345 | * displaying in the explorer/experimenter gui |
---|
346 | */ |
---|
347 | public String classifierTipText() { |
---|
348 | return "The classifier to use for classification."; |
---|
349 | } |
---|
350 | |
---|
351 | /** |
---|
352 | * Sets the classifier to classify instances with. |
---|
353 | * |
---|
354 | * @param value The classifier to be used (with its options set). |
---|
355 | */ |
---|
356 | public void setClassifier(Classifier value) { |
---|
357 | m_Classifier = value; |
---|
358 | } |
---|
359 | |
---|
360 | /** |
---|
361 | * Gets the classifier used by the filter. |
---|
362 | * |
---|
363 | * @return The classifier to be used. |
---|
364 | */ |
---|
365 | public Classifier getClassifier() { |
---|
366 | return m_Classifier; |
---|
367 | } |
---|
368 | |
---|
369 | /** |
---|
370 | * Gets the classifier specification string, which contains the class name of |
---|
371 | * the classifier and any options to the classifier. |
---|
372 | * |
---|
373 | * @return the classifier string. |
---|
374 | */ |
---|
375 | protected String getClassifierSpec() { |
---|
376 | String result; |
---|
377 | Classifier c; |
---|
378 | |
---|
379 | c = getClassifier(); |
---|
380 | result = c.getClass().getName(); |
---|
381 | if (c instanceof OptionHandler) |
---|
382 | result += " " + Utils.joinOptions(((OptionHandler) c).getOptions()); |
---|
383 | |
---|
384 | return result; |
---|
385 | } |
---|
386 | |
---|
387 | /** |
---|
388 | * Returns the tip text for this property |
---|
389 | * |
---|
390 | * @return tip text for this property suitable for |
---|
391 | * displaying in the explorer/experimenter gui |
---|
392 | */ |
---|
393 | public String serializedClassifierFileTipText() { |
---|
394 | return "A file containing the serialized model of a trained classifier."; |
---|
395 | } |
---|
396 | |
---|
397 | /** |
---|
398 | * Gets the file pointing to a serialized, trained classifier. If it is |
---|
399 | * null or pointing to a directory it will not be used. |
---|
400 | * |
---|
401 | * @return the file the serialized, trained classifier is located |
---|
402 | * in |
---|
403 | */ |
---|
404 | public File getSerializedClassifierFile() { |
---|
405 | return m_SerializedClassifierFile; |
---|
406 | } |
---|
407 | |
---|
408 | /** |
---|
409 | * Sets the file pointing to a serialized, trained classifier. If the |
---|
410 | * argument is null, doesn't exist or pointing to a directory, then the |
---|
411 | * value is ignored. |
---|
412 | * |
---|
413 | * @param value the file pointing to the serialized, trained classifier |
---|
414 | */ |
---|
415 | public void setSerializedClassifierFile(File value) { |
---|
416 | if ((value == null) || (!value.exists())) |
---|
417 | value = new File(System.getProperty("user.dir")); |
---|
418 | |
---|
419 | m_SerializedClassifierFile = value; |
---|
420 | } |
---|
421 | |
---|
422 | /** |
---|
423 | * Returns the tip text for this property |
---|
424 | * |
---|
425 | * @return tip text for this property suitable for |
---|
426 | * displaying in the explorer/experimenter gui |
---|
427 | */ |
---|
428 | public String outputClassificationTipText() { |
---|
429 | return "Whether to add an attribute with the actual classification."; |
---|
430 | } |
---|
431 | |
---|
432 | /** |
---|
433 | * Get whether the classifiction of the classifier is output. |
---|
434 | * |
---|
435 | * @return true if the classification of the classifier is output. |
---|
436 | */ |
---|
437 | public boolean getOutputClassification() { |
---|
438 | return m_OutputClassification; |
---|
439 | } |
---|
440 | |
---|
441 | /** |
---|
442 | * Set whether the classification of the classifier is output. |
---|
443 | * |
---|
444 | * @param value whether the classification of the classifier is output. |
---|
445 | */ |
---|
446 | public void setOutputClassification(boolean value) { |
---|
447 | m_OutputClassification = value; |
---|
448 | } |
---|
449 | |
---|
450 | /** |
---|
451 | * Returns the tip text for this property |
---|
452 | * |
---|
453 | * @return tip text for this property suitable for |
---|
454 | * displaying in the explorer/experimenter gui |
---|
455 | */ |
---|
456 | public String removeOldClassTipText() { |
---|
457 | return "Whether to remove the old class attribute."; |
---|
458 | } |
---|
459 | |
---|
460 | /** |
---|
461 | * Get whether the old class attribute is removed. |
---|
462 | * |
---|
463 | * @return true if the old class attribute is removed. |
---|
464 | */ |
---|
465 | public boolean getRemoveOldClass() { |
---|
466 | return m_RemoveOldClass; |
---|
467 | } |
---|
468 | |
---|
469 | /** |
---|
470 | * Set whether the old class attribute is removed. |
---|
471 | * |
---|
472 | * @param value whether the old class attribute is removed. |
---|
473 | */ |
---|
474 | public void setRemoveOldClass(boolean value) { |
---|
475 | m_RemoveOldClass = value; |
---|
476 | } |
---|
477 | |
---|
478 | /** |
---|
479 | * Returns the tip text for this property |
---|
480 | * |
---|
481 | * @return tip text for this property suitable for |
---|
482 | * displaying in the explorer/experimenter gui |
---|
483 | */ |
---|
484 | public String outputDistributionTipText() { |
---|
485 | return |
---|
486 | "Whether to add attributes with the distribution for all classes " |
---|
487 | + "(for numeric classes this will be identical to the attribute output " |
---|
488 | + "with 'outputClassification')."; |
---|
489 | } |
---|
490 | |
---|
491 | /** |
---|
492 | * Get whether the classifiction of the classifier is output. |
---|
493 | * |
---|
494 | * @return true if the distribution of the classifier is output. |
---|
495 | */ |
---|
496 | public boolean getOutputDistribution() { |
---|
497 | return m_OutputDistribution; |
---|
498 | } |
---|
499 | |
---|
500 | /** |
---|
501 | * Set whether the Distribution of the classifier is output. |
---|
502 | * |
---|
503 | * @param value whether the distribution of the classifier is output. |
---|
504 | */ |
---|
505 | public void setOutputDistribution(boolean value) { |
---|
506 | m_OutputDistribution = value; |
---|
507 | } |
---|
508 | |
---|
509 | /** |
---|
510 | * Returns the tip text for this property |
---|
511 | * |
---|
512 | * @return tip text for this property suitable for |
---|
513 | * displaying in the explorer/experimenter gui |
---|
514 | */ |
---|
515 | public String outputErrorFlagTipText() { |
---|
516 | return |
---|
517 | "Whether to add an attribute indicating whether the classifier output " |
---|
518 | + "a wrong classification (for numeric classes this is the numeric " |
---|
519 | + "difference)."; |
---|
520 | } |
---|
521 | |
---|
522 | /** |
---|
523 | * Get whether the classifiction of the classifier is output. |
---|
524 | * |
---|
525 | * @return true if the classification of the classifier is output. |
---|
526 | */ |
---|
527 | public boolean getOutputErrorFlag() { |
---|
528 | return m_OutputErrorFlag; |
---|
529 | } |
---|
530 | |
---|
531 | /** |
---|
532 | * Set whether the classification of the classifier is output. |
---|
533 | * |
---|
534 | * @param value whether the classification of the classifier is output. |
---|
535 | */ |
---|
536 | public void setOutputErrorFlag(boolean value) { |
---|
537 | m_OutputErrorFlag = value; |
---|
538 | } |
---|
539 | |
---|
540 | /** |
---|
541 | * Determines the output format based on the input format and returns |
---|
542 | * this. In case the output format cannot be returned immediately, i.e., |
---|
543 | * immediateOutputFormat() returns false, then this method will be called |
---|
544 | * from batchFinished(). |
---|
545 | * |
---|
546 | * @param inputFormat the input format to base the output format on |
---|
547 | * @return the output format |
---|
548 | * @throws Exception in case the determination goes wrong |
---|
549 | * @see #hasImmediateOutputFormat() |
---|
550 | * @see #batchFinished() |
---|
551 | */ |
---|
552 | protected Instances determineOutputFormat(Instances inputFormat) |
---|
553 | throws Exception { |
---|
554 | |
---|
555 | Instances result; |
---|
556 | FastVector atts; |
---|
557 | int i; |
---|
558 | FastVector values; |
---|
559 | int classindex; |
---|
560 | |
---|
561 | classindex = -1; |
---|
562 | |
---|
563 | // copy old attributes |
---|
564 | atts = new FastVector(); |
---|
565 | for (i = 0; i < inputFormat.numAttributes(); i++) { |
---|
566 | // remove class? |
---|
567 | if ((i == inputFormat.classIndex()) && (getRemoveOldClass()) ) |
---|
568 | continue; |
---|
569 | // record class index |
---|
570 | if (i == inputFormat.classIndex()) |
---|
571 | classindex = i; |
---|
572 | atts.addElement(inputFormat.attribute(i).copy()); |
---|
573 | } |
---|
574 | |
---|
575 | // add new attributes |
---|
576 | // 1. classification? |
---|
577 | if (getOutputClassification()) { |
---|
578 | // if old class got removed, use this one |
---|
579 | if (classindex == -1) |
---|
580 | classindex = atts.size(); |
---|
581 | atts.addElement(inputFormat.classAttribute().copy("classification")); |
---|
582 | } |
---|
583 | |
---|
584 | // 2. distribution? |
---|
585 | if (getOutputDistribution()) { |
---|
586 | if (inputFormat.classAttribute().isNominal()) { |
---|
587 | for (i = 0; i < inputFormat.classAttribute().numValues(); i++) { |
---|
588 | atts.addElement(new Attribute("distribution_" + inputFormat.classAttribute().value(i))); |
---|
589 | } |
---|
590 | } |
---|
591 | else { |
---|
592 | atts.addElement(new Attribute("distribution")); |
---|
593 | } |
---|
594 | } |
---|
595 | |
---|
596 | // 2. error flag? |
---|
597 | if (getOutputErrorFlag()) { |
---|
598 | if (inputFormat.classAttribute().isNominal()) { |
---|
599 | values = new FastVector(); |
---|
600 | values.addElement("no"); |
---|
601 | values.addElement("yes"); |
---|
602 | atts.addElement(new Attribute("error", values)); |
---|
603 | } |
---|
604 | else { |
---|
605 | atts.addElement(new Attribute("error")); |
---|
606 | } |
---|
607 | } |
---|
608 | |
---|
609 | // generate new header |
---|
610 | result = new Instances(inputFormat.relationName(), atts, 0); |
---|
611 | result.setClassIndex(classindex); |
---|
612 | |
---|
613 | return result; |
---|
614 | } |
---|
615 | |
---|
616 | /** |
---|
617 | * Processes the given data (may change the provided dataset) and returns |
---|
618 | * the modified version. This method is called in batchFinished(). |
---|
619 | * |
---|
620 | * @param instances the data to process |
---|
621 | * @return the modified data |
---|
622 | * @throws Exception in case the processing goes wrong |
---|
623 | * @see #batchFinished() |
---|
624 | */ |
---|
625 | protected Instances process(Instances instances) throws Exception { |
---|
626 | Instances result; |
---|
627 | double[] newValues; |
---|
628 | double[] oldValues; |
---|
629 | int i; |
---|
630 | int start; |
---|
631 | int n; |
---|
632 | Instance newInstance; |
---|
633 | Instance oldInstance; |
---|
634 | Instances header; |
---|
635 | double[] distribution; |
---|
636 | File file; |
---|
637 | ObjectInputStream ois; |
---|
638 | |
---|
639 | // load or train classifier |
---|
640 | if (!isFirstBatchDone()) { |
---|
641 | file = getSerializedClassifierFile(); |
---|
642 | if (!file.isDirectory()) { |
---|
643 | ois = new ObjectInputStream(new FileInputStream(file)); |
---|
644 | m_ActualClassifier = (Classifier) ois.readObject(); |
---|
645 | header = null; |
---|
646 | // let's see whether there's an Instances header stored as well |
---|
647 | try { |
---|
648 | header = (Instances) ois.readObject(); |
---|
649 | } |
---|
650 | catch (Exception e) { |
---|
651 | // ignored |
---|
652 | } |
---|
653 | ois.close(); |
---|
654 | // same dataset format? |
---|
655 | if ((header != null) && (!header.equalHeaders(instances))) |
---|
656 | throw new WekaException( |
---|
657 | "Training header of classifier and filter dataset don't match:\n" |
---|
658 | + header.equalHeadersMsg(instances)); |
---|
659 | } |
---|
660 | else { |
---|
661 | m_ActualClassifier = AbstractClassifier.makeCopy(m_Classifier); |
---|
662 | m_ActualClassifier.buildClassifier(instances); |
---|
663 | } |
---|
664 | } |
---|
665 | |
---|
666 | result = getOutputFormat(); |
---|
667 | |
---|
668 | // traverse all instances |
---|
669 | for (i = 0; i < instances.numInstances(); i++) { |
---|
670 | oldInstance = instances.instance(i); |
---|
671 | oldValues = oldInstance.toDoubleArray(); |
---|
672 | newValues = new double[result.numAttributes()]; |
---|
673 | |
---|
674 | start = oldValues.length; |
---|
675 | if (getRemoveOldClass()) |
---|
676 | start--; |
---|
677 | |
---|
678 | // copy old values |
---|
679 | System.arraycopy(oldValues, 0, newValues, 0, start); |
---|
680 | |
---|
681 | // add new values: |
---|
682 | // 1. classification? |
---|
683 | if (getOutputClassification()) { |
---|
684 | newValues[start] = m_ActualClassifier.classifyInstance(oldInstance); |
---|
685 | start++; |
---|
686 | } |
---|
687 | |
---|
688 | // 2. distribution? |
---|
689 | if (getOutputDistribution()) { |
---|
690 | distribution = m_ActualClassifier.distributionForInstance(oldInstance); |
---|
691 | for (n = 0; n < distribution.length; n++) { |
---|
692 | newValues[start] = distribution[n]; |
---|
693 | start++; |
---|
694 | } |
---|
695 | } |
---|
696 | |
---|
697 | // 3. error flag? |
---|
698 | if (getOutputErrorFlag()) { |
---|
699 | if (result.classAttribute().isNominal()) { |
---|
700 | if (oldInstance.classValue() == m_ActualClassifier.classifyInstance(oldInstance)) |
---|
701 | newValues[start] = 0; |
---|
702 | else |
---|
703 | newValues[start] = 1; |
---|
704 | } |
---|
705 | else { |
---|
706 | newValues[start] = m_ActualClassifier.classifyInstance(oldInstance) - oldInstance.classValue(); |
---|
707 | } |
---|
708 | start++; |
---|
709 | } |
---|
710 | |
---|
711 | // create new instance |
---|
712 | if (oldInstance instanceof SparseInstance) |
---|
713 | newInstance = new SparseInstance(oldInstance.weight(), newValues); |
---|
714 | else |
---|
715 | newInstance = new DenseInstance(oldInstance.weight(), newValues); |
---|
716 | |
---|
717 | // copy string/relational values from input to output |
---|
718 | copyValues(newInstance, false, oldInstance.dataset(), getOutputFormat()); |
---|
719 | |
---|
720 | result.add(newInstance); |
---|
721 | } |
---|
722 | |
---|
723 | return result; |
---|
724 | } |
---|
725 | |
---|
726 | /** |
---|
727 | * Returns the revision string. |
---|
728 | * |
---|
729 | * @return the revision |
---|
730 | */ |
---|
731 | public String getRevision() { |
---|
732 | return RevisionUtils.extract("$Revision: 5987 $"); |
---|
733 | } |
---|
734 | |
---|
735 | /** |
---|
736 | * runs the filter with the given arguments |
---|
737 | * |
---|
738 | * @param args the commandline arguments |
---|
739 | */ |
---|
740 | public static void main(String[] args) { |
---|
741 | runFilter(new AddClassification(), args); |
---|
742 | } |
---|
743 | } |
---|