moa.classifiers.lazy.neighboursearch

Class NormalizableDistance

java.lang.Object

moa.classifiers.lazy.neighboursearch.NormalizableDistance

All Implemented Interfaces:

DistanceFunction

Direct Known Subclasses:

EuclideanDistance

public abstract class NormalizableDistance
extends Object
implements DistanceFunction

Represents the abstract ancestor for normalizable distance functions, like Euclidean or Manhattan distance.

Version:

$Revision: 8034 $

Author:

Fracpete (fracpete at waikato dot ac dot nz), Gabi Schmidberger (gabi@cs.waikato.ac.nz) -- original code from weka.core.EuclideanDistance, Ashraf M. Kibriya (amk14@cs.waikato.ac.nz) -- original code from weka.core.EuclideanDistance

Field Summary

Fields

Modifier and Type	Field and Description
protected boolean[]	m_ActiveIndices The boolean flags, whether an attribute will be used or not.
protected weka.core.Instances	m_Data the instances used internally.
protected boolean	m_DontNormalize True if normalization is turned off (default false).
protected double[][]	m_Ranges The range of the attributes.
protected boolean	m_Validated Whether all the necessary preparations have been done.
static int	R_MAX Index in ranges for MAX.
static int	R_MIN Index in ranges for MIN.
static int	R_WIDTH Index in ranges for WIDTH.

Constructor Summary

Constructors

Constructor and Description
NormalizableDistance() Invalidates the distance function, Instances must be still set.
NormalizableDistance(weka.core.Instances data) Initializes the distance function and automatically initializes the ranges.

Method Summary

Methods

Modifier and Type	Method and Description
String	attributeIndicesTipText() Returns the tip text for this property.
protected double	difference(int index, double val1, double val2) Computes the difference between two given attribute values.
double	distance(weka.core.Instance first, weka.core.Instance second) Calculates the distance between two instances.
double	distance(weka.core.Instance first, weka.core.Instance second, double cutOffValue) Calculates the distance between two instances.
String	dontNormalizeTipText() Returns the tip text for this property.
String	getAttributeIndices() Gets the range of attributes used in the calculation of the distance.
boolean	getDontNormalize() Gets whether if the attribute values are to be normazlied in distance calculation.
weka.core.Instances	getInstances() returns the instances currently set.
boolean	getInvertSelection() Gets whether the matching sense of attribute indices is inverted or not.
double[][]	getRanges() Method to get the ranges.
abstract String	globalInfo() Returns a string describing this object.
protected void	initialize() initializes the ranges and the attributes being used.
protected void	initializeAttributeIndices() initializes the attribute indices.
double[][]	initializeRanges() Initializes the ranges using all instances of the dataset.
double[][]	initializeRanges(int[] instList) Initializes the ranges of a subset of the instances of this dataset.
double[][]	initializeRanges(int[] instList, int startIdx, int endIdx) Initializes the ranges of a subset of the instances of this dataset.
void	initializeRangesEmpty(int numAtt, double[][] ranges) Used to initialize the ranges.
boolean	inRanges(weka.core.Instance instance, double[][] ranges) Test if an instance is within the given ranges.
protected void	invalidate() invalidates all initializations.
String	invertSelectionTipText() Returns the tip text for this property.
static boolean	isMissingValue(double val) Tests if the given value codes "missing".
protected double	norm(double x, int i) Normalizes a given value of a numeric attribute.
void	postProcessDistances(double[] distances) Does nothing, derived classes may override it though.
boolean	rangesSet() Check if ranges are set.
void	setAttributeIndices(String value) Sets the range of attributes to use in the calculation of the distance.
void	setDontNormalize(boolean dontNormalize) Sets whether if the attribute values are to be normalized in distance calculation.
void	setInstances(weka.core.Instances insts) Sets the instances.
void	setInvertSelection(boolean value) Sets whether the matching sense of attribute indices is inverted or not.
String	toString() Returns an empty string.
void	update(weka.core.Instance ins) Update the distance function (if necessary) for the newly added instance.
protected abstract double	updateDistance(double currDist, double diff) Updates the current distance calculated so far with the new difference between two attributes.
void	updateRanges(weka.core.Instance instance) Update the ranges if a new instance comes.
double[][]	updateRanges(weka.core.Instance instance, double[][] ranges) Updates the ranges given a new instance.
void	updateRanges(weka.core.Instance instance, int numAtt, double[][] ranges) Updates the minimum and maximum and width values for all the attributes based on a new instance.
void	updateRangesFirst(weka.core.Instance instance, int numAtt, double[][] ranges) Used to initialize the ranges.
protected void	validate() performs the initializations if necessary.

Methods inherited from class java.lang.Object

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

Field Detail

R_MIN

public static final int R_MIN

Index in ranges for MIN.

See Also:

Constant Field Values

R_MAX

public static final int R_MAX

Index in ranges for MAX.

See Also:

Constant Field Values

R_WIDTH

public static final int R_WIDTH

Index in ranges for WIDTH.

See Also:

Constant Field Values

m_Data

protected weka.core.Instances m_Data

the instances used internally.

m_DontNormalize

protected boolean m_DontNormalize

True if normalization is turned off (default false).

m_Ranges

protected double[][] m_Ranges

The range of the attributes.

m_ActiveIndices

protected boolean[] m_ActiveIndices

The boolean flags, whether an attribute will be used or not.

m_Validated

protected boolean m_Validated

Whether all the necessary preparations have been done.

Constructor Detail

NormalizableDistance

public NormalizableDistance()

Invalidates the distance function, Instances must be still set.

NormalizableDistance

public NormalizableDistance(weka.core.Instances data)

Initializes the distance function and automatically initializes the ranges.

Parameters:

data - the instances the distance function should work on

Method Detail

globalInfo

public abstract String globalInfo()

Returns a string describing this object.

Returns:

a description of the evaluator suitable for displaying in the explorer/experimenter gui

dontNormalizeTipText

public String dontNormalizeTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

setDontNormalize

public void setDontNormalize(boolean dontNormalize)

Sets whether if the attribute values are to be normalized in distance calculation.

Parameters:

dontNormalize - if true the values are not normalized

getDontNormalize

public boolean getDontNormalize()

Gets whether if the attribute values are to be normazlied in distance calculation. (default false i.e. attribute values are normalized.)

Returns:

false if values get normalized

attributeIndicesTipText

public String attributeIndicesTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

setAttributeIndices

public void setAttributeIndices(String value)

Sets the range of attributes to use in the calculation of the distance. The indices start from 1, 'first' and 'last' are valid as well. E.g.: first-3,5,6-last

Specified by:

setAttributeIndices in interface DistanceFunction

Parameters:

value - the new attribute index range

getAttributeIndices

public String getAttributeIndices()

Gets the range of attributes used in the calculation of the distance.

Specified by:

getAttributeIndices in interface DistanceFunction

Returns:

the attribute index range

invertSelectionTipText

public String invertSelectionTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

setInvertSelection

public void setInvertSelection(boolean value)

Sets whether the matching sense of attribute indices is inverted or not.

Specified by:

setInvertSelection in interface DistanceFunction

Parameters:

value - if true the matching sense is inverted

getInvertSelection

public boolean getInvertSelection()

Gets whether the matching sense of attribute indices is inverted or not.

Specified by:

getInvertSelection in interface DistanceFunction

Returns:

true if the matching sense is inverted

invalidate

protected void invalidate()

invalidates all initializations.

validate

protected void validate()

performs the initializations if necessary.

initialize

protected void initialize()

initializes the ranges and the attributes being used.

initializeAttributeIndices

protected void initializeAttributeIndices()

initializes the attribute indices.

setInstances

public void setInstances(weka.core.Instances insts)

Sets the instances.

Specified by:

setInstances in interface DistanceFunction

Parameters:

insts - the instances to use

getInstances

public weka.core.Instances getInstances()

returns the instances currently set.

Specified by:

getInstances in interface DistanceFunction

Returns:

the current instances

postProcessDistances

public void postProcessDistances(double[] distances)

Does nothing, derived classes may override it though.

Specified by:

postProcessDistances in interface DistanceFunction

Parameters:

distances - the distances to post-process

update

public void update(weka.core.Instance ins)

Update the distance function (if necessary) for the newly added instance.

Specified by:

update in interface DistanceFunction

Parameters:

ins - the instance to add

distance

public double distance(weka.core.Instance first,
              weka.core.Instance second)

Calculates the distance between two instances.

Specified by:

distance in interface DistanceFunction

Parameters:

first - the first instance

second - the second instance

Returns:

the distance between the two given instances

distance

public double distance(weka.core.Instance first,
              weka.core.Instance second,
              double cutOffValue)

Calculates the distance between two instances. Offers speed up (if the distance function class in use supports it) in nearest neighbour search by taking into account the cutOff or maximum distance. Depending on the distance function class, post processing of the distances by postProcessDistances(double []) may be required if this function is used.

Specified by:

distance in interface DistanceFunction

Parameters:

first - the first instance

second - the second instance

cutOffValue - If the distance being calculated becomes larger than cutOffValue then the rest of the calculation is discarded.

Returns:

the distance between the two given instances or Double.POSITIVE_INFINITY if the distance being calculated becomes larger than cutOffValue.

updateDistance

protected abstract double updateDistance(double currDist,
                    double diff)

Updates the current distance calculated so far with the new difference between two attributes. The difference between the attributes was calculated with the difference(int,double,double) method.

Parameters:

currDist - the current distance calculated so far

diff - the difference between two new attributes

Returns:

the update distance

See Also:

difference(int, double, double)

norm

protected double norm(double x,
          int i)

Normalizes a given value of a numeric attribute.

Parameters:

x - the value to be normalized

i - the attribute's index

Returns:

the normalized value

difference

protected double difference(int index,
                double val1,
                double val2)

Computes the difference between two given attribute values.

Parameters:

index - the attribute index

val1 - the first value

val2 - the second value

Returns:

the difference

initializeRanges

public double[][] initializeRanges()

Initializes the ranges using all instances of the dataset. Sets m_Ranges.

Returns:

the ranges

updateRangesFirst

public void updateRangesFirst(weka.core.Instance instance,
                     int numAtt,
                     double[][] ranges)

Used to initialize the ranges. For this the values of the first instance is used to save time. Sets low and high to the values of the first instance and width to zero.

Parameters:

instance - the new instance

numAtt - number of attributes in the model

ranges - low, high and width values for all attributes

updateRanges

public void updateRanges(weka.core.Instance instance,
                int numAtt,
                double[][] ranges)

Updates the minimum and maximum and width values for all the attributes based on a new instance.

Parameters:

instance - the new instance

numAtt - number of attributes in the model

ranges - low, high and width values for all attributes

initializeRangesEmpty

public void initializeRangesEmpty(int numAtt,
                         double[][] ranges)

Used to initialize the ranges.

Parameters:

numAtt - number of attributes in the model

ranges - low, high and width values for all attributes

updateRanges

public double[][] updateRanges(weka.core.Instance instance,
                      double[][] ranges)

Updates the ranges given a new instance.

Parameters:

instance - the new instance

ranges - low, high and width values for all attributes

Returns:

the updated ranges

initializeRanges

public double[][] initializeRanges(int[] instList)
                            throws Exception

Initializes the ranges of a subset of the instances of this dataset. Therefore m_Ranges is not set.

Parameters:

instList - list of indexes of the subset

Returns:

the ranges

Throws:

Exception - if something goes wrong

initializeRanges

public double[][] initializeRanges(int[] instList,
                          int startIdx,
                          int endIdx)
                            throws Exception

Initializes the ranges of a subset of the instances of this dataset. Therefore m_Ranges is not set. The caller of this method should ensure that the supplied start and end indices are valid (start <= end, end<instList.length etc) and correct.

Parameters:

instList - list of indexes of the instances

startIdx - start index of the subset of instances in the indices array

endIdx - end index of the subset of instances in the indices array

Returns:

the ranges

Throws:

Exception - if something goes wrong

updateRanges

public void updateRanges(weka.core.Instance instance)

Update the ranges if a new instance comes.

Parameters:

instance - the new instance

inRanges

public boolean inRanges(weka.core.Instance instance,
               double[][] ranges)

Test if an instance is within the given ranges.

Parameters:

instance - the instance

ranges - the ranges the instance is tested to be in

Returns:

true if instance is within the ranges

rangesSet

public boolean rangesSet()

Check if ranges are set.

Returns:

true if ranges are set

getRanges

public double[][] getRanges()
                     throws Exception

Method to get the ranges.

Returns:

the ranges

Throws:

Exception - if no randes are set yet

toString

public String toString()

Returns an empty string.

Overrides:

toString in class Object

Returns:

an empty string

isMissingValue

public static boolean isMissingValue(double val)

Tests if the given value codes "missing".

Parameters:

val - the value to be tested

Returns:

true if val codes "missing"