Example usage for org.apache.mahout.classifier.df.data DescriptorUtils parseDescriptor

Introduction

In this page you can find the example usage for org.apache.mahout.classifier.df.data DescriptorUtils parseDescriptor.

Prototype

public static Attribute[] parseDescriptor(CharSequence descriptor) throws DescriptorException 

Source Link

Document

Parses a descriptor string and generates the corresponding array of Attributes

Usage

From source file:com.wsc.myexample.decisionForest.MyDataLoader.java

License:Apache License

/**
 * Generates the Dataset by parsing the entire data
 * //from   www  .java  2 s.com
 * @param descriptor
 *          attributes description
 * @param regression
 *                if true, the label is numerical
 * @param fs
 *          file system
 * @param path
 *          data path
 */
public static MyDataset generateDataset(CharSequence descriptor, boolean regression, String path)
        throws DescriptorException, IOException {
    Attribute[] attrs = DescriptorUtils.parseDescriptor(descriptor);

    Scanner scanner = new Scanner(new File(path));

    // used to convert CATEGORICAL attribute to Integer
    @SuppressWarnings("unchecked")
    Set<String>[] valsets = new Set[attrs.length];

    int size = 0;
    while (scanner.hasNextLine()) {
        String line = scanner.nextLine();
        if (line.isEmpty()) {
            continue;
        }

        if (parseString(attrs, valsets, line, regression)) {
            size++;
        }
    }

    scanner.close();

    @SuppressWarnings("unchecked")
    List<String>[] values = new List[attrs.length];
    for (int i = 0; i < valsets.length; i++) {
        if (valsets[i] != null) {
            values[i] = Lists.newArrayList(valsets[i]);
        }
    }

    return new MyDataset(attrs, values, size, regression);
}

---

From source file:com.wsc.myexample.decisionForest.MyDataLoader.java

License:Apache License

/**
 * Generates the Dataset by parsing the entire data
 * /*w  w w . j av  a 2s . c o  m*/
 * @param descriptor
 *          attributes description
 */
public static MyDataset generateDataset(CharSequence descriptor, boolean regression, String[] data)
        throws DescriptorException {
    Attribute[] attrs = DescriptorUtils.parseDescriptor(descriptor);

    // used to convert CATEGORICAL attributes to Integer
    @SuppressWarnings("unchecked")
    Set<String>[] valsets = new Set[attrs.length];

    int size = 0;
    for (String aData : data) {
        if (aData.isEmpty()) {
            continue;
        }

        if (parseString(attrs, valsets, aData, regression)) {
            size++;
        }
    }

    @SuppressWarnings("unchecked")
    List<String>[] values = new List[attrs.length];
    for (int i = 0; i < valsets.length; i++) {
        if (valsets[i] != null) {
            values[i] = Lists.newArrayList(valsets[i]);
        }
    }

    return new MyDataset(attrs, values, size, regression);
}

---