weka.core
Class SparseInstance

java.lang.Object
  weka.core.AbstractInstance
      weka.core.SparseInstance

All Implemented Interfaces:

Serializable, Copyable, Instance, RevisionHandler

Direct Known Subclasses:

BinarySparseInstance

public class SparseInstance
extends AbstractInstance

Class for storing an instance as a sparse vector. A sparse instance only requires storage for those attribute values that are non-zero. Since the objective is to reduce storage requirements for datasets with large numbers of default values, this also includes nominal attributes -- the first nominal value (i.e. that which has index 0) will not require explicit storage, so rearrange your nominal attribute value orderings if necessary. Missing values will be stored explicitly.

Version:

$Revision: 8034 $

Author:

Eibe Frank

See Also:

Serialized Form

Constructor Summary

SparseInstance(double weight, double[] attValues)
Constructor that generates a sparse instance from the given parameters.

SparseInstance(double weight, double[] attValues, int[] indices, int maxNumValues)
Constructor that inititalizes instance variable with given values.

SparseInstance(Instance instance)
Constructor that generates a sparse instance from the given instance.

SparseInstance(int numAttributes)
Constructor of an instance that sets weight to one, all values to be missing, and the reference to the dataset to null.

SparseInstance(SparseInstance instance)
Constructor that copies the info from the given instance.

Method Summary

Object	copy() Produces a shallow copy of this instance.
String	getRevision() Returns the revision string.
int	index(int position) Returns the index of the attribute stored at the given position.
int	locateIndex(int index) Locates the greatest index that is not greater than the given index.
static void	main(String[] options) Main method for testing this class.
Instance	mergeInstance(Instance inst) Merges this instance with the given instance and returns the result.
int	numAttributes() Returns the number of attributes.
int	numValues() Returns the number of values in the sparse vector.
void	replaceMissingValues(double[] array) Replaces all missing values in the instance with the values contained in the given array.
void	setValue(int attIndex, double value) Sets a specific value in the instance to the given value (internal floating-point format).
void	setValueSparse(int indexOfIndex, double value) Sets a specific value in the instance to the given value (internal floating-point format).
double[]	toDoubleArray() Returns the values of each attribute as an array of doubles.
String	toStringNoWeight() Returns the description of one instance in sparse format.
double	value(int attIndex) Returns an instance's attribute value in internal format.

Methods inherited from class weka.core.AbstractInstance

attribute, attributeSparse, classAttribute, classIndex, classIsMissing, classValue, dataset, deleteAttributeAt, enumerateAttributes, equalHeaders, equalHeadersMsg, hasMissingValue, insertAttributeAt, isMissing, isMissing, isMissingSparse, numClasses, relationalValue, relationalValue, setClassMissing, setClassValue, setClassValue, setDataset, setMissing, setMissing, setValue, setValue, setValue, setWeight, stringValue, stringValue, toString, toString, toString, value, valueSparse, weight

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Constructor Detail

SparseInstance

public SparseInstance(Instance instance)

Constructor that generates a sparse instance from the given instance. Reference to the dataset is set to null. (ie. the instance doesn't have access to information about the attribute types)

Parameters:

instance - the instance from which the attribute values and the weight are to be copied

SparseInstance

public SparseInstance(SparseInstance instance)

Constructor that copies the info from the given instance. Reference to the dataset is set to null. (ie. the instance doesn't have access to information about the attribute types)

Parameters:

instance - the instance from which the attribute info is to be copied

SparseInstance

public SparseInstance(double weight,
                      double[] attValues)

Constructor that generates a sparse instance from the given parameters. Reference to the dataset is set to null. (ie. the instance doesn't have access to information about the attribute types)

Parameters:

weight - the instance's weight

attValues - a vector of attribute values

SparseInstance

public SparseInstance(double weight,
                      double[] attValues,
                      int[] indices,
                      int maxNumValues)

Constructor that inititalizes instance variable with given values. Reference to the dataset is set to null. (ie. the instance doesn't have access to information about the attribute types) Note that the indices need to be sorted in ascending order. Otherwise things won't work properly.

Parameters:

weight - the instance's weight

attValues - a vector of attribute values (just the ones to be stored)

indices - the indices of the given values in the full vector (need to be sorted in ascending order)

maxNumValues - the maximium number of values that can be stored

SparseInstance

public SparseInstance(int numAttributes)

Constructor of an instance that sets weight to one, all values to be missing, and the reference to the dataset to null. (ie. the instance doesn't have access to information about the attribute types)

Parameters:

numAttributes - the size of the instance

Method Detail

copy

public Object copy()

Produces a shallow copy of this instance. The copy has access to the same dataset. (if you want to make a copy that doesn't have access to the dataset, use new SparseInstance(instance)

Returns:

the shallow copy

index

public int index(int position)

Returns the index of the attribute stored at the given position.

Parameters:

position - the position

Returns:

the index of the attribute stored at the given position

locateIndex

public int locateIndex(int index)

Locates the greatest index that is not greater than the given index.

Returns:

the internal index of the attribute index. Returns -1 if no index with this property could be found

mergeInstance

public Instance mergeInstance(Instance inst)

Merges this instance with the given instance and returns the result. Dataset is set to null.

Parameters:

inst - the instance to be merged with this one

Returns:

the merged instances

numAttributes

public int numAttributes()

Returns the number of attributes.

Returns:

the number of attributes as an integer

numValues

public int numValues()

Returns the number of values in the sparse vector.

Returns:

the number of values

replaceMissingValues

public void replaceMissingValues(double[] array)

Replaces all missing values in the instance with the values contained in the given array. A deep copy of the vector of attribute values is performed before the values are replaced.

Parameters:

array - containing the means and modes

Throws:

IllegalArgumentException - if numbers of attributes are unequal

setValue

public void setValue(int attIndex,
                     double value)

Sets a specific value in the instance to the given value (internal floating-point format). Performs a deep copy of the vector of attribute values before the value is set.

Parameters:

attIndex - the attribute's index

value - the new attribute value (If the corresponding attribute is nominal (or a string) then this is the new value's index as a double).

setValueSparse

public void setValueSparse(int indexOfIndex,
                           double value)

Sets a specific value in the instance to the given value (internal floating-point format). Performs a deep copy of the vector of attribute values before the value is set.

Parameters:

indexOfIndex - the index of the attribute's index

value - the new attribute value (If the corresponding attribute is nominal (or a string) then this is the new value's index as a double).

toDoubleArray

public double[] toDoubleArray()

Returns the values of each attribute as an array of doubles.

Returns:

an array containing all the instance attribute values

toStringNoWeight

public String toStringNoWeight()

Returns the description of one instance in sparse format. If the instance doesn't have access to a dataset, it returns the internal floating-point values. Quotes string values that contain whitespace characters.

Returns:

the instance's description as a string

value

public double value(int attIndex)

Returns an instance's attribute value in internal format.

Parameters:

attIndex - the attribute's index

Returns:

the specified value as a double (If the corresponding attribute is nominal (or a string) then it returns the value's index as a double).

main

public static void main(String[] options)

Main method for testing this class.

getRevision

public String getRevision()

Returns the revision string.

Specified by:

getRevision in interface RevisionHandler

Overrides:

getRevision in class AbstractInstance

Returns:

the revision