checkclusterer.java

数据挖掘中聚类的算法

      for (int i = 0; i < train.numInstances(); i++) {
        train.instance(i).setWeight(0);
      }
      Random random = new Random(1);
      for (int i = 0; i < train.numInstances() / 2; i++) {
        int inst = Math.abs(random.nextInt()) % train.numInstances();
        int weight = Math.abs(random.nextInt()) % 10 + 1;
        train.instance(inst).setWeight(weight);
      }
      clusterers[1].buildClusterer(train);
      built = true;
      evaluationI.setClusterer(clusterers[1]);
      if (evaluationB.equals(evaluationI)) {
        //	println("no");
        evalFail = true;
        throw new Exception("evalFail");
      }
      
      println("yes");
      result[0] = true;
    } catch (Exception ex) {
      println("no");
      result[0] = false;
      
      if (m_Debug) {
        println("\n=== Full Report ===");
        
        if (evalFail) {
          println("Results don't differ between non-weighted and "
              + "weighted instance models.");
          println("Here are the results:\n");
          println("\nboth methods\n");
          println(evaluationB.clusterResultsToString());
        } else {
          print("Problem during");
          if (built) {
            print(" testing");
          } else {
            print(" training");
          }
          println(": " + ex.getMessage() + "\n");
        }
        println("Here is the dataset:\n");
        println("=== Train Dataset ===\n"
            + train.toString() + "\n");
        println("=== Train Weights ===\n");
        for (int i = 0; i < train.numInstances(); i++) {
          println(" " + (i + 1) 
              + "    " + train.instance(i).weight());
        }
      }
    }
    
    return result;
  }
  
  /**
   * Checks whether the scheme alters the training dataset during
   * training. If the scheme needs to modify the training
   * data it should take a copy of the training data. Currently checks
   * for changes to header structure, number of instances, order of
   * instances, instance weights.
   *
   * @param nominalPredictor if true use nominal predictor attributes
   * @param numericPredictor if true use numeric predictor attributes
   * @param stringPredictor if true use string predictor attributes
   * @param datePredictor if true use date predictor attributes
   * @param relationalPredictor if true use relational predictor attributes
   * @param multiInstance whether multi-instance is needed
   * @param predictorMissing true if we know the clusterer can handle
   * (at least) moderate missing predictor values
   * @return index 0 is true if the test was passed
   */
  protected boolean[] datasetIntegrity(
      boolean nominalPredictor,
      boolean numericPredictor, 
      boolean stringPredictor, 
      boolean datePredictor,
      boolean relationalPredictor,
      boolean multiInstance,
      boolean predictorMissing) {
    
    print("clusterer doesn't alter original datasets");
    printAttributeSummary(
        nominalPredictor, numericPredictor, stringPredictor, datePredictor, relationalPredictor, multiInstance);
    print("...");
    int numTrain = getNumInstances(), missingLevel = 20;
    
    boolean[] result = new boolean[2];
    Instances train = null;
    Clusterer clusterer = null;
    try {
      train = makeTestDataset(42, numTrain, 
                              nominalPredictor    ? getNumNominal()    : 0,
                              numericPredictor    ? getNumNumeric()    : 0, 
                              stringPredictor     ? getNumString()     : 0, 
                              datePredictor       ? getNumDate()       : 0, 
                              relationalPredictor ? getNumRelational() : 0, 
                              multiInstance);
      if (nominalPredictor && !multiInstance)
        train.deleteAttributeAt(0);
      if (missingLevel > 0)
        addMissing(train, missingLevel, predictorMissing);
      clusterer = Clusterer.makeCopies(getClusterer(), 1)[0];
    } catch (Exception ex) {
      throw new Error("Error setting up for tests: " + ex.getMessage());
    }
    try {
      Instances trainCopy = new Instances(train);
      clusterer.buildClusterer(trainCopy);
      compareDatasets(train, trainCopy);
      
      println("yes");
      result[0] = true;
    } catch (Exception ex) {
      println("no");
      result[0] = false;
      
      if (m_Debug) {
        println("\n=== Full Report ===");
        print("Problem during training");
        println(": " + ex.getMessage() + "\n");
        println("Here is the dataset:\n");
        println("=== Train Dataset ===\n"
            + train.toString() + "\n");
      }
    }
    
    return result;
  }
  
  /**
   * Checks whether an updateable scheme produces the same model when
   * trained incrementally as when batch trained. The model itself
   * cannot be compared, so we compare the evaluation on test data
   * for both models. It is possible to get a false positive on this
   * test (likelihood depends on the classifier).
   *
   * @param nominalPredictor if true use nominal predictor attributes
   * @param numericPredictor if true use numeric predictor attributes
   * @param stringPredictor if true use string predictor attributes
   * @param datePredictor if true use date predictor attributes
   * @param relationalPredictor if true use relational predictor attributes
   * @param multiInstance whether multi-instance is needed
   * @return index 0 is true if the test was passed
   */
  protected boolean[] updatingEquality(
      boolean nominalPredictor,
      boolean numericPredictor, 
      boolean stringPredictor, 
      boolean datePredictor,
      boolean relationalPredictor,
      boolean multiInstance) {
    
    print("incremental training produces the same results"
        + " as batch training");
    printAttributeSummary(
        nominalPredictor, numericPredictor, stringPredictor, datePredictor, relationalPredictor, multiInstance);
    print("...");
    int numTrain = getNumInstances(), missingLevel = 0;
    boolean predictorMissing = false, classMissing = false;
    
    boolean[] result = new boolean[2];
    Instances train = null;
    Clusterer[] clusterers = null;
    ClusterEvaluation evaluationB = null;
    ClusterEvaluation evaluationI = null;
    boolean built = false;
    try {
      train = makeTestDataset(42, numTrain, 
                              nominalPredictor    ? getNumNominal()    : 0,
                              numericPredictor    ? getNumNumeric()    : 0, 
                              stringPredictor     ? getNumString()     : 0, 
                              datePredictor       ? getNumDate()       : 0, 
                              relationalPredictor ? getNumRelational() : 0, 
                              multiInstance);
      if (missingLevel > 0)
        addMissing(train, missingLevel, predictorMissing, classMissing);
      clusterers = Clusterer.makeCopies(getClusterer(), 2);
      evaluationB = new ClusterEvaluation();
      evaluationI = new ClusterEvaluation();
      clusterers[0].buildClusterer(train);
      evaluationB.setClusterer(clusterers[0]);
    } catch (Exception ex) {
      throw new Error("Error setting up for tests: " + ex.getMessage());
    }
    try {
      clusterers[1].buildClusterer(new Instances(train, 0));
      for (int i = 0; i < train.numInstances(); i++) {
        ((UpdateableClusterer)clusterers[1]).updateClusterer(
            train.instance(i));
      }
      built = true;
      evaluationI.setClusterer(clusterers[1]);
      if (!evaluationB.equals(evaluationI)) {
        println("no");
        result[0] = false;
        
        if (m_Debug) {
          println("\n=== Full Report ===");
          println("Results differ between batch and "
              + "incrementally built models.\n"
              + "Depending on the classifier, this may be OK");
          println("Here are the results:\n");
          println("\nbatch built results\n" + evaluationB.clusterResultsToString());
          println("\nincrementally built results\n" + evaluationI.clusterResultsToString());
          println("Here are the datasets:\n");
          println("=== Train Dataset ===\n"
              + train.toString() + "\n");
        }
      }
      else {
        println("yes");
        result[0] = true;
      }
    } catch (Exception ex) {
      result[0] = false;
      
      print("Problem during");
      if (built)
        print(" testing");
      else
        print(" training");
      println(": " + ex.getMessage() + "\n");
    }
    
    return result;
  }
  
  /**
   * Runs a text on the datasets with the given characteristics.
   * 
   * @param nominalPredictor if true use nominal predictor attributes
   * @param numericPredictor if true use numeric predictor attributes
   * @param stringPredictor if true use string predictor attributes
   * @param datePredictor if true use date predictor attributes
   * @param relationalPredictor if true use relational predictor attributes
   * @param multiInstance whether multi-instance is needed
   * @param missingLevel the percentage of missing values
   * @param predictorMissing true if the missing values may be in 
   * the predictors
   * @param numTrain the number of instances in the training set
   * @param accepts the acceptable string in an exception
   * @return index 0 is true if the test was passed, index 1 is true if test 
   *         was acceptable
   */
  protected boolean[] runBasicTest(boolean nominalPredictor,
      boolean numericPredictor, 
      boolean stringPredictor,
      boolean datePredictor,
      boolean relationalPredictor,
      boolean multiInstance,
      int missingLevel,
      boolean predictorMissing,
      int numTrain,
      FastVector accepts) {
    
    boolean[] result = new boolean[2];
    Instances train = null;
    Clusterer clusterer = null;
    try {
      train = makeTestDataset(42, numTrain, 
                              nominalPredictor    ? getNumNominal()    : 0,
                              numericPredictor    ? getNumNumeric()    : 0, 
                              stringPredictor     ? getNumString()     : 0,
                              datePredictor       ? getNumDate()       : 0,
                              relationalPredictor ? getNumRelational() : 0,
                              multiInstance);
      if (nominalPredictor && !multiInstance)
        train.deleteAttributeAt(0);
      if (missingLevel > 0)
        addMissing(train, missingLevel, predictorMissing);
      clusterer = Clusterer.makeCopies(getClusterer(), 1)[0];
    } catch (Exception ex) {
      ex.printStackTrace();
      throw new Error("Error setting up for tests: " + ex.getMessage());
    }
    try {
      clusterer.buildClusterer(train);
      println("yes");
      result[0] = true;
    } 
    catch (Exception ex) {
      boolean acceptable = false;
      String msg = ex.getMessage().toLowerCase();
      for (int i = 0; i < accepts.size(); i++) {
        if (msg.indexOf((String)accepts.elementAt(i)) >= 0) {
          acceptable = true;
        }
      }
      
      println("no" + (acceptable ? " (OK error message)" : ""));
      result[1] = acceptable;
      
      if (m_Debug) {
        println("\n=== Full Report ===");
        print("Problem during training");
        println(": " + ex.getMessage() + "\n");
        if (!acceptable) {
          if (accepts.size() > 0) {
            print("Error message doesn't mention ");
            for (int i = 0; i < accepts.size(); i++) {
              if (i != 0) {
                print(" or ");
              }
              print('"' + (String)accepts.elementAt(i) + '"');
            }
          }
          println("here is the dataset:\n");
          println("=== Train Dataset ===\n"
              + train.toString() + "\n");
        }
      }
    }
    
    return result;
  }
  
  /**
   * Add missing values to a dataset.
   *
   * @param data the instances to add missing values to
   * @param level the level of missing values to add (if positive, this
   * is the probability that a value will be set to missing, if negative
   * all but one value will be set to missing (not yet implemented))
   * @param predictorMissing if true, predictor attributes will be modified
   */
  protected void addMissing(Instances data, int level, boolean predictorMissing) {
    
    Random random = new Random(1);
    for (int i = 0; i < data.numInstances(); i++) {
      Instance current = data.instance(i);
      for (int j = 0; j < data.numAttributes(); j++) {
        if (predictorMissing) {
          if (Math.abs(random.nextInt()) % 100 < level)
            current.setMissing(j);
        }
      }
    }
  }
  
  /**
   * Make a simple set of instances with variable position of the class 
   * attribute, which can later be modified for use in specific tests.
   *
   * @param seed the random number seed
   * @param numInstances the number of instances to generate
   * @param numNominal the number of nominal attributes
   * @param numNumeric the number of numeric attributes
   * @param numString the number of string attributes
   * @param numDate the number of date attributes
   * @param numRelational the number of relational attributes
   * @param multiInstance whether the dataset should a multi-instance dataset
   * @return the test dataset
   * @throws Exception if the dataset couldn't be generated
   * @see TestInstances#CLASS_IS_LAST
   */
  protected Instances makeTestDataset(int seed, int numInstances, 
                                      int numNominal, int numNumeric, 
                                      int numString, int numDate,
                                      int numRelational,
                                      boolean multiInstance)
  throws Exception {
    
    TestInstances dataset = new TestInstances();
    
    dataset.setSeed(seed);
    dataset.setNumInstances(numInstances);
    dataset.setNumNominal(numNominal);
    dataset.setNumNumeric(numNumeric);
    dataset.setNumString(numString);
    dataset.setNumDate(numDate);
    dataset.setNumRelational(numRelational);
    dataset.setClassIndex(TestInstances.NO_CLASS);
    dataset.setMultiInstance(multiInstance);
    
    return dataset.generate();
  }
  
  /**
   * Print out a short summary string for the dataset characteristics
   *
   * @param nominalPredictor true if nominal predictor attributes are present
   * @param numericPredictor true if numeric predictor attributes are present
   * @param stringPredictor true if string predictor attributes are present
   * @param datePredictor true if date predictor attributes are present
   * @param relationalPredictor true if relational predictor attributes are present
   * @param multiInstance whether multi-instance is needed
   */
  protected void printAttributeSummary(boolean nominalPredictor, 
                                       boolean numericPredictor, 
                                       boolean stringPredictor, 
                                       boolean datePredictor, 
                                       boolean relationalPredictor, 
                                       boolean multiInstance) {
    
    String str = "";

    if (numericPredictor)
      str += "numeric";
    
    if (nominalPredictor) {
      if (str.length() > 0)
        str += " & ";
      str += "nominal";
    }
    
    if (stringPredictor) {
      if (str.length() > 0)
        str += " & ";
      str += "string";
    }
    
    if (datePredictor) {
      if (str.length() > 0)
        str += " & ";
      str += "date";
    }
    
    if (relationalPredictor) {
      if (str.length() > 0)
        str += " & ";
      str += "relational";
    }
    
    str = " (" + str + " predictors)";
    
    print(str);
  }
  
  /**
   * Test method for this class
   * 
   * @param args the commandline options
   */
  public static void main(String [] args) {
    runCheck(new CheckClusterer(), args);
  }
}