Package org.tribuo.util.infotheory.example

Class InformationTheoryDemo

java.lang.Object

org.tribuo.util.infotheory.example.InformationTheoryDemo

public class InformationTheoryDemo
extends Object

Demo showing how to calculate various mutual informations and entropies.

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	InformationTheoryDemo.DemoOptions	Command line options.
static enum	InformationTheoryDemo.DistributionType	Type of data distribution.

Constructor Summary

Constructors

Constructor	Description
InformationTheoryDemo()

Method Summary

Modifier and Type	Method	Description
static CachedTriple<List<Integer>,List<Integer>,List<Integer>>	generateCorrelated(int length, int alphabetSize, double xyCorrelation, double xzCorrelation)	These correlations don't map to mutual information values, as if xyDraw is above xyCorrelation then the draw is completely random.
static List<Integer>	generateUniform(int length, int alphabetSize)	Generates a sample from a uniform distribution over the integers.
static CachedTriple<List<Integer>,List<Integer>,List<Integer>>	generateXOR(int length)	Generates a sample from a three variable XOR function.
static void	main(String[] args)	Runs a simple demo of the information theory functions.

Methods inherited from class java.lang.Object

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

Constructor Details

InformationTheoryDemo

public InformationTheoryDemo()

Method Details

generateUniform

public static List<Integer> generateUniform(int length,
 int alphabetSize)

Generates a sample from a uniform distribution over the integers.

Parameters:

length - The number of samples.

alphabetSize - The alphabet size (i.e., the number of unique values).

Returns:

A sample from a uniform distribution.

generateXOR

public static CachedTriple<List<Integer>,List<Integer>,List<Integer>> generateXOR(int length)

Generates a sample from a three variable XOR function.

Each list is a binary variable, and the third is the XOR of the first two.

Parameters:

length - The number of samples.

Returns:

A sample from an XOR function.

generateCorrelated

public static CachedTriple<List<Integer>,List<Integer>,List<Integer>> generateCorrelated(int length,
 int alphabetSize,
 double xyCorrelation,
 double xzCorrelation)

These correlations don't map to mutual information values, as if xyDraw is above xyCorrelation then the draw is completely random.

To make it generate correlations of a specific mutual information then it needs to specify the full joint distribution and draw from that.

Parameters:

length - The number of samples.

alphabetSize - The alphabet size (i.e., the number of unique values).

xyCorrelation - Value between 0.0 and 1.0 specifying how likely it is that Y has the same value as X.

xzCorrelation - Value between 0.0 and 1.0 specifying how likely it is that Z has the same value as X.

Returns:

A triple of samples drawn from correlated random variables.

main

public static void main(String[] args)

Runs a simple demo of the information theory functions.

Parameters:

args - The CLI arguments.