Java ClusterEvaluation.setClusterer方法代码示例

import weka.clusterers.ClusterEvaluation; //导入方法依赖的package包/类
private int guessNumClusters(EM clusterer, Instances instances, int start, int end) throws Exception {
    ClusterEvaluation eval = new ClusterEvaluation();
    int bestNum = start;
    double best = Double.POSITIVE_INFINITY;
    double bic;
    for (int c = start; c <= end; c++) {
        clusterer.setNumClusters(c);
        clusterer.buildClusterer(instances);
        eval.setClusterer(clusterer);
        eval.evaluateClusterer(instances);
        bic = bic(eval.getLogLikelihood(), c, instances.numInstances());
        logger.trace("numCluster " + c + " -> BIC: " + bic);
        if (bic < best) {
            best = bic;
            bestNum = c;
            logger.trace("bestNum: " + bestNum);
        }
    }
    return bestNum;
}
 

import weka.clusterers.ClusterEvaluation; //导入方法依赖的package包/类
public void generateClassToCluster(){
	Remove filter = new Remove();
	filter.setAttributeIndices("" + (weather.classIndex() + 1));
	try {
		filter.setInputFormat(weather);
		Instances dataClusterer = Filter.useFilter(weather, filter);
		clusterer = new EM();
		clusterer.buildClusterer(dataClusterer);
		ClusterEvaluation eval = new ClusterEvaluation();
		eval.setClusterer(clusterer);
		eval.evaluateClusterer(weather);

		System.out.println(eval.clusterResultsToString());
	} catch (Exception e) {
	}
}
 

import weka.clusterers.ClusterEvaluation; //导入方法依赖的package包/类
/**
  * Gets the results for the supplied train and test datasets.
  *
  * @param train the training Instances.
  * @param test the testing Instances.
  * @return the results stored in an array. The objects stored in
  * the array may be Strings, Doubles, or null (for the missing value).
  * @exception Exception if a problem occurs while getting the results
  */
 public Object [] getResult(Instances train, Instances test) 
   throws Exception {
   
   if (m_clusterer == null) {
     throw new Exception("No clusterer has been specified");
   }
   int addm = (m_additionalMeasures != null) 
     ? m_additionalMeasures.length 
     : 0;
   int overall_length = RESULT_SIZE+addm;

   if (m_removeClassColumn && train.classIndex() != -1) {
     // remove the class column from the training and testing data
     Remove r = new Remove();
     r.setAttributeIndicesArray(new int [] {train.classIndex()});
     r.setInvertSelection(false);
     r.setInputFormat(train);
     train = Filter.useFilter(train, r);
     
     test = Filter.useFilter(test, r);
   }
   train.setClassIndex(-1);
   test.setClassIndex(-1);
     

   ClusterEvaluation eval = new ClusterEvaluation();

   Object [] result = new Object[overall_length];
   long trainTimeStart = System.currentTimeMillis();
   m_clusterer.buildClusterer(train);
   double numClusters = m_clusterer.numberOfClusters();
   eval.setClusterer(m_clusterer);
   long trainTimeElapsed = System.currentTimeMillis() - trainTimeStart;
   long testTimeStart = System.currentTimeMillis();
   eval.evaluateClusterer(test);
   long testTimeElapsed = System.currentTimeMillis() - testTimeStart;
   //    m_result = eval.toSummaryString();

   // The results stored are all per instance -- can be multiplied by the
   // number of instances to get absolute numbers
   int current = 0;
   result[current++] = new Double(train.numInstances());
   result[current++] = new Double(test.numInstances());

   result[current++] = new Double(eval.getLogLikelihood());
   result[current++] = new Double(numClusters);
   
   // Timing stats
   result[current++] = new Double(trainTimeElapsed / 1000.0);
   result[current++] = new Double(testTimeElapsed / 1000.0);
   
   for (int i=0;i<addm;i++) {
     if (m_doesProduce[i]) {
try {
  double dv = ((AdditionalMeasureProducer)m_clusterer).
    getMeasure(m_additionalMeasures[i]);
  Double value = new Double(dv);
  
  result[current++] = value;
} catch (Exception ex) {
  System.err.println(ex);
}
     } else {
result[current++] = null;
     }
   }
   
   if (current != overall_length) {
     throw new Error("Results didn't fit RESULT_SIZE");
   }
   return result;
 }
 

import weka.clusterers.ClusterEvaluation; //导入方法依赖的package包/类
@Override
public void start() throws Exception {
    if (clusterer == null || data == null) {
        throw new Exception("Please provide data examples and clusterer.");
    }

    // Przygotuj dane -> Ustaw klasę decyzyjną
    if (data.classIndex() == -1) {
        data.setClassIndex(data.numAttributes() - 1);
    }

    // Rozdziel przykłady na dwie klasy: mniejszościową i większościową
    Instances minorityInstances = new Instances(data);
    Instances majorityInstances = new Instances(data);
    separateDecisionClasses(data, minorityInstances, majorityInstances);

    // Pokaż statystyki przykladow
    showInitialDataStatitistics(minorityInstances, majorityInstances);

    // Usuń klasę decyzyjną -> uczenie nienadzorowane
    weka.filters.unsupervised.attribute.Remove filter = new weka.filters.unsupervised.attribute.Remove();
    filter.setAttributeIndices("" + (minorityInstances.classIndex() + 1));
    filter.setInputFormat(minorityInstances);
    Instances minorityInstancesNoClass = Filter.useFilter(minorityInstances, filter);
    Instances majorityInstancesNoClass = Filter.useFilter(majorityInstances, filter);
    Instances dataNoClass = Filter.useFilter(data, filter);

    // Stwórz skupiska

    DBSCAN dbScan = new DBSCAN();
    dbScan.setDatabase_Type("weka.clusterers.forOPTICSAndDBScan.Databases.SequentialDatabase");
    dbScan.setDatabase_distanceType("weka.clusterers.forOPTICSAndDBScan.DataObjects.EuclideanDataObject");
    dbScan.setEpsilon(0.5);
    dbScan.setMinPoints(4);
    clusterer = dbScan;
    clusterer.buildClusterer(minorityInstancesNoClass);

    // Pokaż wyniki analizy skupisk
    ClusterEvaluation evaluation = new ClusterEvaluation();
    evaluation.setClusterer(clusterer);
    evaluation.evaluateClusterer(minorityInstancesNoClass);
    System.out.println(evaluation.clusterResultsToString());

    // Klasyfikuj przyklady klasy wiekszosciowej do skupisk (nie modyfukuje skupisk)
    Map<Integer, Integer> majorityHistogram = createHistogram(clusterer, majorityInstancesNoClass);

    Map<Integer, Integer> minorityHistogram = createHistogram(clusterer, minorityInstancesNoClass);

    // Pokaz statystyki
    showHistogram(majorityHistogram, "Majority");
    showHistogram(minorityHistogram, "Minority");
    showCombinedHistograms(majorityHistogram, minorityHistogram);

    Map<Integer, Instances> clustersAssigmentsMap = new TreeMap<Integer, Instances>();


}
 

import weka.clusterers.ClusterEvaluation; //导入方法依赖的package包/类
@Override
public void start() throws Exception {
    if (clusterer == null || data == null) {
        throw new Exception("Please provide data examples and clusterer.");
    }

    // Przygotuj dane -> Ustaw klasę decyzyjną
    if (data.classIndex() == -1) {
        data.setClassIndex(data.numAttributes() - 1);
    }

    // Rozdziel przykłady na dwie klasy: mniejszościową i większościową
    Instances minorityInstances = new Instances(data);
    Instances majorityInstances = new Instances(data);
    separateDecisionClasses(data, minorityInstances, majorityInstances);

    // Pokaż statystyki przykladow
    showInitialDataStatitistics(data, minorityInstances, majorityInstances);

    // Usuń klasę decyzyjną -> uczenie nienadzorowane
    weka.filters.unsupervised.attribute.Remove filter = new weka.filters.unsupervised.attribute.Remove();
    filter.setAttributeIndices("" + (minorityInstances.classIndex() + 1));
    filter.setInputFormat(minorityInstances);
    Instances minorityInstancesNoClass = Filter.useFilter(minorityInstances, filter);
    Instances majorityInstancesNoClass = Filter.useFilter(majorityInstances, filter);
    Instances dataNoClass = Filter.useFilter(data, filter);

    // Stwórz skupiska
    String options;
    if (autoParametrizationEnable) {
        options = getParametersForKMeans(minorityInstancesNoClass, filename);
        clusterer.setOptions(weka.core.Utils.splitOptions(options));
    }

    clusterer.buildClusterer(minorityInstancesNoClass);

    // Pokaż wyniki analizy skupisk
    ClusterEvaluation evaluation = new ClusterEvaluation();
    evaluation.setClusterer(clusterer);
    evaluation.evaluateClusterer(minorityInstancesNoClass);
    System.out.println(evaluation.clusterResultsToString());

    // Klasyfikuj przyklady klasy wiekszosciowej do skupisk (nie modyfukuje skupisk)
    Map<Integer, Integer> majorityHistogram = createHistogram(clusterer, majorityInstancesNoClass);

    Map<Integer, Integer> minorityHistogram = createHistogram(clusterer, minorityInstancesNoClass);

    // Pokaz statystyki
    showHistogram(majorityHistogram, "Majority");
    showHistogram(minorityHistogram, "Minority");
    showCombinedHistograms(majorityHistogram, minorityHistogram);

    // Rozdziel przyklady z klasami decyzyjnymi wedlug wyznaczonych skupisk
    Map<Integer, Instances> clustersAssignmentsMap = createClusterAssignmentsMap(data, dataNoClass, clusterer);
    showClusterAssignmentsMap(clustersAssignmentsMap);

    //SMOTE + Klasyfikacja
    System.out.println("Classifiers");
    for (Map.Entry<Integer, Instances> entry : clustersAssignmentsMap.entrySet()) {


        FilteredClassifier fc = new FilteredClassifier();
        if (smoteEnable) {
            System.out.println("SMOTE filtering");
            SMOTE smote = new SMOTE();
            smote.setOptions(weka.core.Utils.splitOptions("-C 0 -K 5 -P " + getSMOTEPercentage(entry.getValue()) + " -S 1"));
            smote.setInputFormat(entry.getValue());
            fc.setFilter(smote);
        }
        fc.setClassifier(classifier);
        Evaluation eval = new Evaluation(entry.getValue());
        eval.crossValidateModel(fc, entry.getValue(), 10, new Random(1));

        System.out.println(eval.toSummaryString("\nResults\n======\n", false));
        System.out.println(eval.toMatrixString());

    }


}
 

import weka.clusterers.ClusterEvaluation; //导入方法依赖的package包/类
public static String getParametersForKMeans(Instances instances, String name) {

        SimpleKMeans kmeans = new SimpleKMeans();
        String options = "-init 0 -max-candidates 100 -periodic-pruning 10000 -min-density 2.0 -t1 -1.25 -t2 -1.0 -V -N 2 -A \"weka.core.EuclideanDistance -R first-last\" -I 500 -num-slots 1 -S 10";
        try {
            kmeans.setOptions(weka.core.Utils.splitOptions(options));
            int numK = 20;
            double[] results = new double[numK];

            for (int i = 1; i <= numK; i++) {
                kmeans.setNumClusters(i);
                kmeans.buildClusterer(instances);


                double[][] centroids = new double[i][kmeans.getClusterCentroids().get(0).numAttributes()];


                double[][] stdDevs = new double[i][kmeans.getClusterStandardDevs().get(0).numAttributes()];

                for (int j = 0; j < i; j++) {
                    for (int k = 0; k < kmeans.getClusterCentroids().get(0).numAttributes(); k++) {
                        if (!Double.isNaN(kmeans.getClusterCentroids().get(j).value(k)) && !Double.isNaN(kmeans.getClusterStandardDevs().get(j).value(k))) {
                            centroids[j][k] = kmeans.getClusterCentroids().get(j).value(k);
                            stdDevs[j][k] = kmeans.getClusterStandardDevs().get(j).value(k);
                        } else {
                            centroids[j][k] = 0.0;
                            stdDevs[j][k] = 0.0;
                        }

                    }
                }


                double sum_clust = 0;
                for (int j = 0; j < centroids.length; j++) {
                    double sum_att = 0;
                    for (int k = 0; k < centroids[0].length; k++) {
                        double d = stdDevs[j][k];
                        sum_att += d;
                    }
                    sum_clust += sum_att;
                }
                results[i - 1] = sum_clust / centroids.length;

                // Pokaż wyniki analizy skupisk
                ClusterEvaluation evaluation = new ClusterEvaluation();
                evaluation.setClusterer(kmeans);
                evaluation.evaluateClusterer(instances);

                //System.out.println(evaluation.clusterResultsToString());
            }
            System.out.println(name);
            for (int i = 0; i < numK; i++) {
                System.out.println(String.format("%.3f", results[i]));
            }
            System.out.println();
        } catch (Exception e) {
            e.printStackTrace();
        }


        return "";

    }
 

import weka.clusterers.ClusterEvaluation; //导入方法依赖的package包/类
public static void main(String[] args) {
    try {

        ConverterUtils.DataSource source = new ConverterUtils.DataSource("fertility_Diagnosis.arff");
        Instances instances = source.getDataSet();
        instances.deleteAttributeAt(9);



        SelfOrganizingMap som = new SelfOrganizingMap();

        som.setConvergenceEpochs(1000);

        som.setLearningRate(0.3);
        som.setOrderingEpochs(50);


        som.setWidth(3);
        som.setCalcStats(true);

        som.buildClusterer(instances);






        ClusterEvaluation eval = new ClusterEvaluation();


//
//
        eval.setClusterer(som);
        eval.evaluateClusterer(instances);
        System.out.println(eval.clusterResultsToString());

        for (int i=0;i<instances.size();i++){
            double[] dist=som.distributionForInstance(instances.get(i));
//            System.out.println(Utils.arrayToString(dist));
        }

        //
//        System.out.println(eval.clusterResultsToString());
//        System.out.println(som.toString());


//        System.out.println(som.toString());










    }catch (Exception ignored){
        ignored.printStackTrace();
    }

}