edu.udo.cs.yale.example
Class FastExample2SparseTransform

java.lang.Object
  edu.udo.cs.yale.example.FastExample2SparseTransform

public class FastExample2SparseTransform
extends java.lang.Object

This class can be used for the efficient generation of sparse example formats. The constructor creates a mapping between example set and example table attribute indices which only need to be performed once. The sparse data can then be queried efficiently by using the methods getNonDefaultAttributeIndices(Example) and getNonDefaultAttributeValues(Example). Please note that this filter should be reinstatiated for new example sets. Furthermore, a gain in performance is only achieved for examples with underlying DoubleSparseArrayDataRows.

Version:

$Id: FastExample2SparseTransform.java,v 2.2 2006/03/21 15:35:39 ingomierswa Exp $

Author:

Julien Nioche, Ingo Mierswa

Field Summary

private int[]

mapping The mapping between the attribute indices in the data row / example table on the attribute indices of the given example set.

Constructor Summary

FastExample2SparseTransform(ExampleSet es)
Returns for a table giving the equivalence between the positions of the Attributes in the ExampleTable and the number of the regular Attributes in the ExampleSet.

Method Summary

int[]	getNonDefaultAttributeIndices(Example example) Returns a list with the indices of the regular Attributes with non-default values.
double[]	getNonDefaultAttributeValues(Example example) Returns an array of non-default values of the given example.
double[]	getNonDefaultAttributeValues(Example example, int[] nonDefaultIndices) Returns an array of non-default values of the given example.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

mapping

private int[] mapping

The mapping between the attribute indices in the data row / example table on the attribute indices of the given example set. This mapping is only necessary for examples backed up by DoubleSparseArrayDataRows.

Constructor Detail

FastExample2SparseTransform

public FastExample2SparseTransform(ExampleSet es)

Returns for a table giving the equivalence between the positions of the Attributes in the ExampleTable and the number of the regular Attributes in the ExampleSet. A value of -1 indicates that the Attribute is not regular or has been deleted. This is used in order to optimize the access to sparse DataRows (e.g. SVM implementations), which is important when the number of Attributes is large.

Method Detail

getNonDefaultAttributeIndices

public int[] getNonDefaultAttributeIndices(Example example)

Returns a list with the indices of the regular Attributes with non-default values. This can be used for a faster construction of sparse dataset representations when the number of Attributes is large. The positions of attributes are sorted by ascending number.

getNonDefaultAttributeValues

public double[] getNonDefaultAttributeValues(Example example)

Returns an array of non-default values of the given example. These are only the values of regular attributes. Simpley invokes getNonDefaultAttributeValues(Example, int[]) with the array of non-default indices for the given example.

getNonDefaultAttributeValues

public double[] getNonDefaultAttributeValues(Example example,
                                             int[] nonDefaultIndices)

Returns an array of non-default values of the given example. These are only the values of regular attributes. The size of the returned array is the same as the size of the given indices array.