de.unidu.is.statistics
Class DiscreteDistribution

java.lang.Object
  de.unidu.is.statistics.Distribution
      de.unidu.is.statistics.DiscreteDistribution

public class DiscreteDistribution

extends Distribution

An class modelling discrete (empirical) probabilistic distributions.

The original values from which the distribution is derived are sorted into bins of equal size. The first bin starts at the minimum value, the last one at the maximum value. This allows for using this distribution for plotting.

If the distribution should be used for computing actual values, the original values are used.

Since:

2002-04-09

Version:

$Revision: 1.4 $, $Date: 2005/02/21 17:29:27 $

Author:

Henrik Nottelmann

Field Summary

protected double	max Maximum original value.
protected double	max2 Maximum original value s.th. the integral from 0 to this value if less than 0.999.
protected double	min Minimum original value.
protected static int	parts Number of bins minus 1.
protected double[]	values Original values from which the distribution is derived.
protected double[]	x The x values (bin starting points) of the distribution when plotted.
protected double[]	y The y values (bin frequencies) of the distribution when plotted.

Constructor Summary

DiscreteDistribution(double[] values)
Constructs a new distribution, based on the specified original values.

Method Summary

void	computeValues(int num, double[] ret) Computes actual values based on the underlying distribution when having the specified number of items, and writes them in decreasing order into the specified array (as many values as possible).
void	eliminateZero() Eliminates info about the distribution at zero.
double	getMax() Returns the maximum original value.
double	getMin() Returns the minimum original value.
double[]	getX() Returns an array with the x values (bin starting points) of the distribution when plotted.
double[]	getY() Returns an array with the y values (bin frequencies) of the distribution when plotted.

Methods inherited from class de.unidu.is.statistics.Distribution

computeValues

Methods inherited from class java.lang.Object

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

Field Detail

x

protected double[] x

The x values (bin starting points) of the distribution when plotted.

y

protected double[] y

The y values (bin frequencies) of the distribution when plotted.

parts

protected static final int parts

Number of bins minus 1.

See Also:

Constant Field Values

values

protected double[] values

Original values from which the distribution is derived.

min

protected double min

Minimum original value.

max

protected double max

Maximum original value.

max2

protected double max2

Maximum original value s.th. the integral from 0 to this value if less than 0.999.

Constructor Detail

DiscreteDistribution

public DiscreteDistribution(double[] values)

Constructs a new distribution, based on the specified original values.

This constructor also creates the bins. The first bin starts at min, the last one at max.

Parameters:

values - original values from which the distribution is derived

Method Detail

getX

public double[] getX()

Returns an array with the x values (bin starting points) of the distribution when plotted.

Returns:

array with the x values (bin starting points)

getY

public double[] getY()

Returns an array with the y values (bin frequencies) of the distribution when plotted.

Returns:

array with the y values (bin frequencies)

getMin

public double getMin()

Returns the minimum original value.

Returns:

minimum original value

getMax

public double getMax()

Returns the maximum original value.

Returns:

maximum original value

eliminateZero

public void eliminateZero()

Eliminates info about the distribution at zero. This is only useful for discrete distributions with a high peak at zero, as then a plot might look better.

Overrides:

eliminateZero in class Distribution

computeValues

public void computeValues(int num,
                          double[] ret)

Computes actual values based on the underlying distribution when having the specified number of items, and writes them in decreasing order into the specified array (as many values as possible).

This method uses the original values and repeates them num/values.lentgh times. Thus, num >= values.lentgh must hold!

Specified by:

computeValues in class Distribution

Parameters:

num - number of items in total

ret - array for the actual values