weka.estimators
Class KernelEstimator

java.lang.Object
  extended byweka.estimators.KernelEstimator
All Implemented Interfaces:
Estimator, java.io.Serializable

public class KernelEstimator
extends java.lang.Object
implements Estimator

Simple kernel density estimator. Uses one gaussian kernel per observed data value.

Version:
$Revision: 1.5 $
Author:
Len Trigg (trigg@cs.waikato.ac.nz)
See Also:
Serialized Form

Field Summary
private  boolean m_AllWeightsOne
          Whether we can optimise the kernel summation
private  int m_NumValues
          Number of values stored in m_Weights and m_Values so far
private  double m_Precision
          The precision of data values
private  double m_StandardDev
          The standard deviation
private  double m_SumOfWeights
          The sum of the weights so far
private  double[] m_Values
          Vector containing all of the values seen
private  double[] m_Weights
          Vector containing the associated weights
private static double MAX_ERROR
          Maximum percentage error permitted in probability calculations
 
Constructor Summary
KernelEstimator(double precision)
          Constructor that takes a precision argument.
 
Method Summary
 void addValue(double data, double weight)
          Add a new data value to the current estimator.
private  int findNearestValue(double key)
          Execute a binary search to locate the nearest data value
 double getProbability(double data)
          Get a probability estimate for a value.
static void main(java.lang.String[] argv)
          Main method for testing this class.
private  double round(double data)
          Round a data value using the defined precision for this estimator
 java.lang.String toString()
          Display a representation of this estimator
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
 

Field Detail

m_Values

private double[] m_Values
Vector containing all of the values seen


m_Weights

private double[] m_Weights
Vector containing the associated weights


m_NumValues

private int m_NumValues
Number of values stored in m_Weights and m_Values so far


m_SumOfWeights

private double m_SumOfWeights
The sum of the weights so far


m_StandardDev

private double m_StandardDev
The standard deviation


m_Precision

private double m_Precision
The precision of data values


m_AllWeightsOne

private boolean m_AllWeightsOne
Whether we can optimise the kernel summation


MAX_ERROR

private static double MAX_ERROR
Maximum percentage error permitted in probability calculations

Constructor Detail

KernelEstimator

public KernelEstimator(double precision)
Constructor that takes a precision argument.

Parameters:
precision - the precision to which numeric values are given. For example, if the precision is stated to be 0.1, the values in the interval (0.25,0.35] are all treated as 0.3.
Method Detail

findNearestValue

private int findNearestValue(double key)
Execute a binary search to locate the nearest data value

Returns:
the index of the nearest data value

round

private double round(double data)
Round a data value using the defined precision for this estimator

Parameters:
data - the value to round
Returns:
the rounded data value

addValue

public void addValue(double data,
                     double weight)
Add a new data value to the current estimator.

Specified by:
addValue in interface Estimator
Parameters:
data - the new data value
weight - the weight assigned to the data value

getProbability

public double getProbability(double data)
Get a probability estimate for a value.

Specified by:
getProbability in interface Estimator
Parameters:
data - the value to estimate the probability of
Returns:
the estimated probability of the supplied value

toString

public java.lang.String toString()
Display a representation of this estimator


main

public static void main(java.lang.String[] argv)
Main method for testing this class.

Parameters:
argv - should contain a sequence of numeric values