Example usage for weka.classifiers.misc InputMappedClassifier setModelHeader

Introduction

In this page you can find the example usage for weka.classifiers.misc InputMappedClassifier setModelHeader.

Prototype

public void setModelHeader(Instances modelHeader) 

Source Link

Document

Set the structure of the data used to create the model.

Usage

From source file:fcul.viegas.ml.learners.NetworkStreamLearningClassifierMapFunction.java

public void open(Configuration parameters) throws Exception {

    try {/*from   w w  w  . j  av a  2s .com*/
        BufferedReader reader = new BufferedReader(new FileReader(NetworkPacketsDefinitions.arffPath));
        ArffLoader.ArffReader arff = new ArffLoader.ArffReader(reader);
        weka.core.Instances dataTrain = arff.getData();
        dataTrain.setClassIndex(dataTrain.numAttributes() - 1);

        Random rand = new Random();
        rand.setSeed(this.hashCode() ^ 83484856847L);
        ArrayList<Integer> listaFeaturesChoosen = new ArrayList<>();
        for (int j = 1; j < dataTrain.numAttributes(); j++) {
            listaFeaturesChoosen.add(j);
        }

        //chose features to remove
        int numberOfFeaturesToBeRemoved = (int) (listaFeaturesChoosen.size()
                * (1.0f - percentOfSubsetFeatures));
        for (int j = 0; j < numberOfFeaturesToBeRemoved; j++) {
            int numberOfFeature = rand.nextInt(listaFeaturesChoosen.size());
            listaFeaturesChoosen.remove(numberOfFeature);
        }
        System.out.println("LEARNING: featureSubset: " + NetworkStreamLearningClassifierMapFunction
                .printFeatures(listaFeaturesChoosen, dataTrain.numAttributes() - 1));
        mappingFeatures = "featureSubset: " + NetworkStreamLearningClassifierMapFunction
                .printFeatures(listaFeaturesChoosen, dataTrain.numAttributes() - 1);
        FastVector attributes = new FastVector();
        int indexLista = 0;
        for (int i = 0; i < NetworkPacketsDefinitions.numberOfFeatures; i++) {
            if (indexLista < listaFeaturesChoosen.size() && listaFeaturesChoosen.get(indexLista) == (i + 1)) {
                indexLista++;
                attributes.addElement(new Attribute("att" + (i + 1)));
            }
        }
        FastVector classLabels = new FastVector();
        for (int i = 0; i < NetworkPacketsDefinitions.numberOfClasses; i++) {
            classLabels.addElement("class" + (i + 1));
        }
        attributes.addElement(new Attribute("class", classLabels));

        this.streamLearningInstanceHeader = new InstancesHeader(new Instances("FlinkPhd", attributes, 0));

        this.streamLearningInstanceHeader.setClassIndex(attributes.size());

        listaFeaturesChoosen.add(dataTrain.numAttributes());

        String[] options = new String[2];
        options[0] = "-R";

        String optRemove = "";
        for (int j = 0; j < listaFeaturesChoosen.size() - 1; j++) {
            optRemove = optRemove + listaFeaturesChoosen.get(j) + ",";
        }
        optRemove = optRemove + listaFeaturesChoosen.get(listaFeaturesChoosen.size() - 1);
        options[1] = optRemove;

        Remove remove = new Remove();
        remove.setOptions(options);
        remove.setInvertSelection(true);
        remove.setInputFormat(dataTrain);

        weka.core.Instances newdataFeat = Filter.useFilter(dataTrain, remove);

        HoeffdingTree tree = new HoeffdingTree();

        weka.classifiers.misc.InputMappedClassifier classifier = new weka.classifiers.misc.InputMappedClassifier();
        classifier.setModelHeader(newdataFeat);
        classifier.setClassifier(tree);
        classifier.buildClassifier(newdataFeat);
        this.classifier = classifier;
        this.coreInstances = dataTrain;
        this.coreInstances.clear();

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

---

From source file:fcul.viegas.ml.learners.NetworkStreamLearningClassifierUpdateMapFunction.java

public void open(Configuration parameters) throws Exception {

    try {//from   ww w  . j a  v  a2 s  .c  o  m
        BufferedReader reader = new BufferedReader(new FileReader(NetworkPacketsDefinitions.arffPath));
        ArffLoader.ArffReader arff = new ArffLoader.ArffReader(reader);
        weka.core.Instances dataTrain = arff.getData();
        dataTrain.setClassIndex(dataTrain.numAttributes() - 1);

        Random rand = new Random();
        rand.setSeed(this.hashCode() ^ 83484856847L);
        ArrayList<Integer> listaFeaturesChoosen = new ArrayList<>();
        for (int j = 1; j < dataTrain.numAttributes(); j++) {
            listaFeaturesChoosen.add(j);
        }

        //chose features to remove
        int numberOfFeaturesToBeRemoved = (int) (listaFeaturesChoosen.size()
                * (1.0f - percentOfSubsetFeatures));
        for (int j = 0; j < numberOfFeaturesToBeRemoved; j++) {
            int numberOfFeature = rand.nextInt(listaFeaturesChoosen.size());
            listaFeaturesChoosen.remove(numberOfFeature);
        }
        System.out.println("UPDATES featureSubset: " + NetworkStreamLearningClassifierMapFunction
                .printFeatures(listaFeaturesChoosen, dataTrain.numAttributes() - 1));
        mappingFeatures = "featureSubset: " + NetworkStreamLearningClassifierMapFunction
                .printFeatures(listaFeaturesChoosen, dataTrain.numAttributes() - 1);
        FastVector attributes = new FastVector();
        int indexLista = 0;
        for (int i = 0; i < NetworkPacketsDefinitions.numberOfFeatures; i++) {
            if (indexLista < listaFeaturesChoosen.size() && listaFeaturesChoosen.get(indexLista) == (i + 1)) {
                indexLista++;
                attributes.addElement(new Attribute("att" + (i + 1)));
            }
        }
        FastVector classLabels = new FastVector();
        for (int i = 0; i < NetworkPacketsDefinitions.numberOfClasses; i++) {
            classLabels.addElement("class" + (i + 1));
        }
        attributes.addElement(new Attribute("class", classLabels));

        this.streamLearningInstanceHeader = new InstancesHeader(new Instances("FlinkPhd", attributes, 0));

        this.streamLearningInstanceHeader.setClassIndex(attributes.size());

        listaFeaturesChoosen.add(dataTrain.numAttributes());

        String[] options = new String[2];
        options[0] = "-R";

        String optRemove = "";
        for (int j = 0; j < listaFeaturesChoosen.size() - 1; j++) {
            optRemove = optRemove + listaFeaturesChoosen.get(j) + ",";
        }
        optRemove = optRemove + listaFeaturesChoosen.get(listaFeaturesChoosen.size() - 1);
        options[1] = optRemove;

        Remove remove = new Remove();
        remove.setOptions(options);
        remove.setInvertSelection(true);
        remove.setInputFormat(dataTrain);

        weka.core.Instances newdataFeat = Filter.useFilter(dataTrain, remove);

        HoeffdingTree tree = new HoeffdingTree();

        weka.classifiers.misc.InputMappedClassifier classifier = new weka.classifiers.misc.InputMappedClassifier();
        classifier.setModelHeader(newdataFeat);
        classifier.setClassifier(tree);
        classifier.buildClassifier(newdataFeat);
        this.classifier = classifier;
        this.coreInstances = dataTrain;

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

---