conceptdriftsimulator.java

著名的开源仿真软件yale

	while (tokenizer.hasMoreTokens() && (streamIndex < noOfStreams)) {
	    streamNames[streamIndex] = tokenizer.nextToken();
	    outString += "  " + streamNames[streamIndex];
	    streamIndex++;
	}
	LogService.logMessage ("ConceptDriftSimulator '"+getName()+"':  " + outString,
			       LogService.MINIMUM);
	return streamNames;
    }

    /** return the index of a stream name in the stream names array. */
    private int  getStreamIndex (String[] streamNames, String streamName) {
	for (int i=0; i < streamNames.length; i++) {
	    if (streamName.equals(streamNames[i])) { return i; }
	}
	LogService.logMessage("Unknown stream name: "+streamName+". Stream names are: "+Arrays.asList(streamNames), LogService.ERROR);
	
	return ILLEGAL_STREAM_INDEX;   // given stream name not found in the stream name array
    }
    /** return the index of a stream name */
    private int  getStreamIndex (String streamName) {
	return getStreamIndex (this.streamNames, streamName);
    }



    /** scan the relevance specification for the data streams from the given string. */
    private double[][]  scanStreamRelevanceSpecification (double[][] streamRelevance, String streamRelevanceString,
							  String[] streamNames, ExampleSet inputSet) throws OperatorException {
	if (streamRelevanceString == null) {
    	    throw new FatalException(getName()+": Parameter 'data_stream_relevance' not set");
	}
	int  streamIndex = 0;
	int  batchIndex  = 0;
	try {
	    BufferedReader  reader      = new BufferedReader(new StringReader(streamRelevanceString));
	    String          line        = null;
	    StringTokenizer tokenizer   = null;
	    String          streamName  = null;
	    String          token       = null;
	    Attribute       label       = inputSet.getLabel();

	    // loop to scan drift specification for each given topic:
	    while ((line = reader.readLine()) != null) {
		line      = line.trim();
		tokenizer = new StringTokenizer (line);
		if (tokenizer.hasMoreTokens()) {
		    streamName  = tokenizer.nextToken();
		    streamIndex = getStreamIndex(streamNames, streamName);
		    if (streamIndex == ILLEGAL_STREAM_INDEX) {
			throw new FatalException("ConceptDriftSimulator '" + getName() + "':  " +
						  "illegal stream name '"+streamName+"' read in 'data_stream_relevance', " +
						  "which did not occur in 'data_stream_names'");
		    }
		    if (tokenizer.hasMoreTokens()) {
			// skip ':' after stream name:
			tokenizer.nextToken();                 
			batchIndex = 0;
			// loop to retrieve and set relevance values for that topic, batch by batch:
			while (tokenizer.hasMoreTokens() && (batchIndex < noOfBatches)) {
			    token = tokenizer.nextToken();
			    streamRelevance[streamIndex][batchIndex] = Double.parseDouble(token);
			    batchIndex++;
			}
		    }
		}
	    }
	} catch (Exception e) {
	    throw new FatalException("ConceptDriftSimulator '" + getName() + "':  "+
				     "Exception while reading the 'data_stream_relevance' values:", e);
	}

     	// ---- print relevance matrix to log file ----
	String  relevanceMatrixString = "\nRelevance probability of each stream over time:\n";
	for (streamIndex=0; streamIndex < noOfStreams; streamIndex++) {
	    relevanceMatrixString += "  " + streamNames[streamIndex] + ":";
	    for (batchIndex=0; batchIndex < noOfBatches; batchIndex++) {
		relevanceMatrixString += "  " + streamRelevance[streamIndex][batchIndex];
	    }
	    relevanceMatrixString += "\n";
	}
	LogService.logMessage ("ConceptDriftSimulator '"+getName()+"':  "+relevanceMatrixString,
			       LogService.TASK);
	return streamRelevance;
    }



    /** perform sampling: randomly assign examples to batches (distribute example stream as evenly as
     *  possible to the batches)
     */
    private void  sampleExamples2Batches () {
	ExampleReader  exampleIterator;
	Example        currentExample;
	int     e;                   // index of the current example
	int     s;                   // index of the stream that the current current examples comes from
	int     t;                   // time index of the current example
	int     b;                   // index of the batch that the current example from stream 's' should be assigned to
	String  currentStreamName;   // name of the stream (e.g. topic) the current example (e.g. document) belongs to

	// maps an example (e.g. document) to a time point and vice versa:
	example2timeIndex  = new int[noOfExamples];
	timeIndex2example  = new int[noOfExamples];
	// maps an example (e.g. document) to a batch:
	example2batchIndex = new int[noOfExamples];

	// noOfStreamExamplesInBatch[s][b] = no. of examples from stream 's' in batch 'b':
	noOfStreamExamplesInBatch    = new int[noOfStreams][noOfBatches];  
	// noOfExamplesInBatch[b] = no. of examples in batch 'b':
	noOfExamplesInBatch          = new int[noOfBatches];
	// currentBatchForStream[s] = batch that the current example from stream 's' should be assigned to:
	int[]  currentBatchForStream = new int[noOfStreams];
	// example2streamIndex[e] = index of the stream the example 'e' comes from
	example2streamIndex          = new int[noOfExamples];

	// ---- initialize all mappings ----
	for (e = 0; e < noOfExamples; e++) {
	    example2streamIndex[e] = 0;
	    example2timeIndex[e]   = 0;
	    example2batchIndex[e]  = 0;
	}
	for (s = 0; s < noOfStreams; s++) {
	    for (b = 0; b < noOfBatches; b++)
		noOfStreamExamplesInBatch[s][b] = 0;
	    currentBatchForStream[s] = 0;
	}
	for (b = 0; b < noOfBatches; b++)
	    noOfExamplesInBatch[b] = 0;

	// ---- create 'example2timeIndex' (mapping of example index to time index) ----
	LogService.logMessage ("ConceptDriftSimulator '" + getName() + "': sample mapping example2timeIndex.",
			       LogService.TASK);
	for (int i=0; i < noOfExamples; i++)  timeIndex2example[i] = i;
	scrambleArray (timeIndex2example);
	// for (int i=0; i < noOfExamples; i++)  example2timeIndex[i] = i;
	// scrambleArray (example2timeIndex);

	// ---- create 'example2batchIndex' (mapping of example index to batch index) ----
	// -- first loop:  retrieve 'example2stream' mapping and initialize weights and user interest --
	// LogService.logMessage ("ConceptDriftSimulator '" + getName() + "': sample mapping example2batchIndex (step 1).",
	//		          LogService.TASK);
	exampleIterator = exampleSet.getExampleReader();  // example iterator
	e = 0;                                            // example index
	while ((e < noOfExamples) && (exampleIterator.hasNext())) {                          // for each example:
	    currentExample = exampleIterator.next();                                         // * get the current example
	    // LogService.logMessage ("ConceptDriftSimulator '" + getName() + "': ... \n" +
	    //		              "  streamNameAttribute = " + streamNameAttribute.getName() +
	    // 		              "  example index = " + e + ".\n", LogService.TASK);
	    currentStreamName = currentExample.getValueAsString (streamNameAttribute);       // * get its stream name
	    s = getStreamIndex(currentStreamName);                                           // * get its stream index
	    example2streamIndex[e] = s;                                                      // * store its stream index
	    try {
		currentExample.setWeight (1.0);                                              // * set its weight to 1.0
	    } catch (MethodNotSupportedException exception) {
		throw new RuntimeException("ConceptDriftSimulator '"+getName()+"':  " +
				       "weight attribute must exist to set weights " +
				       "(internal error, should never occur).");
	    }
	    currentExample.setLabel (Attribute.NEGATIVE_CLASS);                              // * initialize its user interest
	    // try {
	    //   currentExample.setValue (userInterestAttribute, Attribute.NEGATIVE_CLASS);  // * initialize its user interest
	    // } catch (MethodNotSupportedException exception) {
	    // 	 throw new RuntimeException("ConceptDriftSimulator '"+getName()+"':  " +
	    //			       "user interest attribute (= new class label)" +
	    //			       " must be nominal (internal error, should " +
	    //			       "never occur).");
	    // }
	    e++;                                                                             // * increment example index
	} // -- end of first loop --
	//
	// -- second loop:  create 'example2timeIndex' and 'example2batchIndex' mapping --
	// LogService.logMessage ("ConceptDriftSimulator '" + getName() + "': sample mapping example2batchIndex (step 2).",
	//		          LogService.TASK);
	t = 0;
	while (t < (noOfBatches * batchSize)) {                                        // for each time point:
	    e = timeIndex2example[t];                                                  // * get its example's index
	    example2timeIndex[e] = t;                                                  // * set example2timeIndex mapping
	    s = example2streamIndex[e];                                                // * get its example's stream index
	    b = currentBatchForStream[s];                                              // * get its example's batch index
	    if (noOfExamplesInBatch[b] < batchSize) {                                  //
		example2batchIndex[e] = b;                                             // * set its example's batch index
		t++;                                                                   // * increment time point
		noOfStreamExamplesInBatch[s][b]++;                                     // * counter for log file output
	    }                                                                          //
	    currentBatchForStream[s] = (currentBatchForStream[s] + 1) % noOfBatches;   // * modulo increment batch index
	}
	while (t < noOfExamples) {                                                     // for each remaining time point:
	    e = timeIndex2example[t];                                                  // * get its example's index
	    example2timeIndex[e] = t;                                                  // * set example2timeIndex mapping
	    example2batchIndex[e] = noOfBatches;                                       // * set its example's batch index
	    t++;                                                                       //   to a value outside simulation
	} // -- end of second loop --
	//
	// -- third loop:  set time index and batch index --
	// LogService.logMessage ("ConceptDriftSimulator '" + getName() + "': sample mapping example2batchIndex (step 3).",
	//		          LogService.TASK);
	exampleIterator = exampleSet.getExampleReader();  // example iterator
	e = 0;                                            // example index
	while ((e < noOfExamples) && (exampleIterator.hasNext())) {                         // for each example:
	    currentExample = exampleIterator.next();                                        // * get the current example
	    currentExample.setValue (timeIndexAttribute, (double)(example2timeIndex[e]));   // * set its time index
	    currentExample.setValue (batchIndexAttribute, (double)(example2batchIndex[e])); // * set its batch index
	    if (example2batchIndex[e] == noOfBatches) {                                     // * if example is outside simu-
		try {                                                                       //   lation, set its weight to 0.0
		    currentExample.setWeight (0.0);                                         //
		} catch (MethodNotSupportedException exception) {
		    throw new RuntimeException("ConceptDriftSimulator '"+getName()+"':  " +
					       "weight attribute must exist to set weights " +
					       "(internal error, should never occur).");                           
		}                                                                      //
	    }                                                                          //
	    e++;                                                                       // * increment example index
	} // -- end of third loop --

	// ---- log file output: stream 2 batch distribution ----
	String  output = "\n  Number of documents of each stream in each batch:\n";
	output += "    Batch:       ";
	for (b = 0; b < noOfBatches; b++) {
	    output += "   " + b;
	}
	output += "\n";
       	for (s = 0; s < noOfStreams; s++) {