AbstractMetric

Overview

Package

Class

Use

Tree

Deprecated

Index

Help

PREV CLASS NEXT CLASS

FRAMES NO FRAMES

SUMMARY: NESTED | FIELD | CONSTR | METHOD

DETAIL: FIELD | CONSTR | METHOD

at.tuwien.ifs.somtoolbox.layers.metrics
Class AbstractMetric

java.lang.Object
  at.tuwien.ifs.somtoolbox.layers.metrics.AbstractMetric

All Implemented Interfaces:: DistanceMetric, Comparable<DistanceMetric>

Direct Known Subclasses:: AbstractWeightedMetric, CosineMetric, L1Metric, L2Metric, L2MetricSparse, LInfinityMetric, LnAlphaMetric, LnMetric, MahalanobisMetric, MissingValueMetricWrapper

public abstract class AbstractMetric
extends Object
implements DistanceMetric
extends Object
implements DistanceMetric

Implements a static method for metric instantiation and a method for mean vector calculation. Furthermore, implements the convenience methods to call the abstract method having two double arrays as arguments (see distance(double[], double[])). This method has to be implemented by classes actually implementing a certain metric.

Version:: $Id: AbstractMetric.java 3883 2010-11-02 17:13:23Z frank $
Author:: Michael Dittenbach, Rudolf Mayer

Nested Class Summary

Nested classes/interfaces inherited from interface at.tuwien.ifs.somtoolbox.layers.metrics.DistanceMetric
`DistanceMetric.SparcseMetricModes`

Field Summary
`static String`	`PACKAGE_NAME`

Constructor Summary
`AbstractMetric()`

Method Summary
`protected void`	`checkDimensions(double[] vector1, double[] vector2)` Performs a check on wether the given vectors have the same dimension.
`protected void`	`checkDimensions(cern.colt.matrix.DoubleMatrix1D vector1, cern.colt.matrix.DoubleMatrix1D vector2)` Performs a check on wether the given vectors have the same dimension.
`int`	`compareTo(DistanceMetric o)`
`abstract double`	`distance(double[] vector1, double[] vector2)` Calculates the distance between two vectors provided by argument `vector1` and `vector2`.
`double`	`distance(double[] vector1, cern.colt.matrix.DoubleMatrix1D vector2)` Calculates the distance between two vectors provided by argument `vector1` and `vector2`.
`double`	`distance(double[] vector, InputDatum data)` Calculates the distance between two vectors provided by argument `vector` and `datum`.
`double`	`distance(cern.colt.matrix.DoubleMatrix1D vector1, double[] vector2)` Calculates the distance between two vectors provided by argument `vector1` and `vector2`.
`double`	`distance(cern.colt.matrix.DoubleMatrix1D vector1, cern.colt.matrix.DoubleMatrix1D vector2)` Calculates the distance between two vectors provided by argument `vector1` and `vector2`.
`double`	`distance(cern.colt.matrix.DoubleMatrix1D vector, InputDatum datum)` Calculates the distance between two vectors provided by argument `vector` and `datum`.
`double`	`distance(InputDatum data, double[] vector)` Calculates the distance between two vectors provided by argument `datum` and `vector`.
`double`	`distance(InputDatum datum, cern.colt.matrix.DoubleMatrix1D vector)` Calculates the distance between two vectors provided by argument `datum` and `vector`.
`double`	`distance(InputDatum datum, InputDatum datum2)` Calculates the distance between two vectors provided by argument `datum` and `datum2`.
`static DistanceMetric`	`instantiate(String mName)` Instantiates a certain distance metric class specified by argument `mName`. Note: for backwards compatibility, if the metric name contains the package `prefix at.ec3.somtoolbox`, this will be replaced by `at.tuwien.ifs.somtoolbox`.
`static DistanceMetric`	`instantiateNice(String metricName)` Same as `instantiate(String)`, but tries to get the metric with the specified name, and then with the package prefix, and throwing only a `SOMToolboxException` with the root cause nested.
`static double[]`	`meanVector(double[] vector1, double[] vector2)` Calculates the mean vector of two double array vectors.
`protected static void`	`performanceTest(DistanceMetric metric, int dim)` Can be used to do some performance testing to compare colt vs.
`void`	`setMetricParams(String metricParamString)` Empty implementation, subclasses needing to set parameters have to override this class.
`double`	`transformValue(double value)`
`double[]`	`transformVector(double[] vector)`

Methods inherited from class java.lang.Object
`clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait`

Field Detail

PACKAGE_NAME

public static final String PACKAGE_NAME

Constructor Detail

AbstractMetric

public AbstractMetric()

Method Detail

instantiate

public static DistanceMetric instantiate(String mName)
                                  throws ClassNotFoundException,
                                         InstantiationException,
                                         IllegalAccessException

Instantiates a certain distance metric class specified by argument mName.
Note: for backwards compatibility, if the metric name contains the package prefix at.ec3.somtoolbox, this will be replaced by at.tuwien.ifs.somtoolbox.

Parameters:: mName - the name of the metric.
Returns:: a distance metric object of class mName.
Throws:: ClassNotFoundException - if class denoted by argument mName is not found.; InstantiationException - if if this Class represents an abstract class, an interface, an array class, a primitive type, or void; or if the class has no nullary constructor; or if the instantiation fails for some other reason.; IllegalAccessException - if the class or its nullary constructor is not accessible.

instantiateNice

public static DistanceMetric instantiateNice(String metricName)
                                      throws SOMToolboxException

Same as instantiate(String), but tries to get the metric with the specified name, and then with the package prefix, and throwing only a SOMToolboxException with the root cause nested.

Throws:: SOMToolboxException

meanVector

public static double[] meanVector(double[] vector1,
                                  double[] vector2)
                           throws MetricException

Calculates the mean vector of two double array vectors.

Parameters:: vector1 - first vector.; vector2 - second vector.
Returns:: the mean vector.
Throws:: MetricException - if the dimensionalities of the two vectors differ.

performanceTest

protected static void performanceTest(DistanceMetric metric,
                                      int dim)

Can be used to do some performance testing to compare colt vs. direct distance implementations.