Creating a simple sparse ARFF file

Here you will find a sample java program to generate sparse ARFF file. The following italic text text has been taken from: http://www.cs.waikato.ac.nz/ml/weka/arff.html

Sparse ARFF files are very similar to ARFF files, but data with value 0 are not be explicitly represented.Sparse ARFF files have the same header (i.e @relation and @attribute tags) but the data section is different. Instead of representing each value in order, like this:

@data

0, X, 0, Y, "class A"

0, 0, W, 0, "class B"

the non-zero attributes are explicitly identified by attribute number and their value stated, like this:

@data

{1 X, 3 Y, 4 "class A"}

{2 W, 4 "class B"}

Each instance is surrounded by curly braces, and the format for each entry is: <index> <space> <value> where index is the attribute index (starting from 0).

The following Java code generates a sparse Instances object and outputs it to stdout and a physical file as ARFF file.

            1: import java.io.File;

           2: import weka.core.Attribute;

           3: import weka.core.FastVector;

           4: import weka.core.Instance;

           5: import weka.core.Instances;

           6: import weka.core.converters.ArffSaver;

           7: import weka.filters.Filter;

           8: import weka.filters.unsupervised.instance.NonSparseToSparse;

           9: /**

          10: * Generates a sparse ARFF file.

          11: *

          12: * @author Razan

          13: */

          14: public class AttTest {

          15:  public static void main(String[] args) throws Exception 

          16: {

          17:     FastVector attributes;

          18:     Instances dataSet;

          19:     double[] values;

          20:     attributes = new FastVector();

          21:      

          22:     attributes.addElement(new Attribute("att1")); 

          23:     attributes.addElement(new Attribute("att2")); 

          24:     attributes.addElement(new Attribute("att3")); 

          25:     attributes.addElement(new Attribute("att4"));

          26:     

          27:     dataSet = new Instances("ESDN", attributes, 0);

          28:      

          29:     values = new double[dataSet.numAttributes()]; 

          30:     values[0] = 3;

          31:     values[1] =7;

          32:     values[3] = 1;

          33:     dataSet.add(new Instance(1.0, values));

          34:     

          35:     values = new double[dataSet.numAttributes()]; 

          36:     values[2] = 2;

          37:     values[3] = 8;

          38:     dataSet.add(new Instance(1.0, values));

          39:     

          40:     NonSparseToSparse nonSparseToSparseInstance = new NonSparseToSparse(); 

          41:     nonSparseToSparseInstance.setInputFormat(dataSet); 

          42:     Instances sparseDataset = Filter.useFilter(dataSet, nonSparseToSparseInstance);

          43:      

          44:     System.out.println(sparseDataset);

          45:     

          46:     ArffSaver arffSaverInstance = new ArffSaver(); 

          47:     arffSaverInstance.setInstances(sparseDataset); 

          48:     arffSaverInstance.setFile(new File("ESDN.arff")); 

          49:     arffSaverInstance.writeBatch();

          50:  }

          51: }

The output of the sample program is in the following:

Hope this will save some of your time.

Technorati Tags: WEKA,RapidMiner,JAVA

Posted on Tuesday, November 8, 2011 3:14 AM WEKA , RapidMiner , JAVA | Back to top