com.aliasi.stats

Class BernoulliEstimator

java.lang.Object

com.aliasi.stats.AbstractDiscreteDistribution

com.aliasi.stats.MultivariateDistribution

com.aliasi.stats.BernoulliDistribution

com.aliasi.stats.BernoulliEstimator

All Implemented Interfaces:

DiscreteDistribution, Compilable

public class BernoulliEstimator
extends BernoulliDistribution
implements Compilable

A BernoulliEstimator provides a maximum likelihood estimate of a Bernoulli distribution. Training samples are provided through the method train(boolean,int) specifying success or failure and the number of samples. An unbiased estimator for a Bernoulli distribution's probability of success is simply the percentage of successes.

Since:

LingPipe2.0

Version:

2.4

Author:

Bob Carpenter

Field Summary

Fields inherited from class com.aliasi.stats.BernoulliDistribution

FAILURE_LABEL, SUCCESS_LABEL

Constructor Summary

Constructors

Constructor and Description
BernoulliEstimator() Construct a Bernoulli estimator with zero counts.

Method Summary

Methods

Modifier and Type	Method and Description
void	compileTo(ObjectOutput objOut) Compiles this Bernoulli estimator to the specified object output.
long	numTrainingSamples() Returns the number of training samples provided for this estimator.
long	numTrainingSamples(boolean success) Returns the number of training samples for the specified outcome, success or failure.
double	successProbability() Returns the maximum likelihood estimate of success from the training samples provided.
void	train(boolean success) Trains the estimator with one sample that is specified as a success or failure.
void	train(boolean success, int numSamples) Train this estimator with the specified number of samples for success or failure as specified.

Methods inherited from class com.aliasi.stats.BernoulliDistribution

label, maxOutcome, numDimensions, probability, variance

Methods inherited from class com.aliasi.stats.MultivariateDistribution

log2Probability, minOutcome, outcome, probability

Methods inherited from class com.aliasi.stats.AbstractDiscreteDistribution

cumulativeProbability, cumulativeProbabilityGreater, cumulativeProbabilityLess, entropy, log2Probability, mean

Methods inherited from class java.lang.Object

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

Constructor Detail

BernoulliEstimator

public BernoulliEstimator()

Construct a Bernoulli estimator with zero counts.

Method Detail

train

public void train(boolean success,
         int numSamples)

Train this estimator with the specified number of samples for success or failure as specified.

Parameters:

success - A flag for whether the training samples are for success or failure.

numSamples - Number of samples to train.

train

public void train(boolean success)

Trains the estimator with one sample that is specified as a success or failure.

Parameters:

success - Flag for whether the sample was a success or a failure.

successProbability

public double successProbability()

Returns the maximum likelihood estimate of success from the training samples provided.

Specified by:

successProbability in class BernoulliDistribution

Returns:

The maximum likelihood estimate of success from the training samples provided.

numTrainingSamples

public long numTrainingSamples()

Returns the number of training samples provided for this estimator.

Returns:

The number of training samples provided for this estimator.

numTrainingSamples

public long numTrainingSamples(boolean success)

Returns the number of training samples for the specified outcome, success or failure.

Parameters:

success - Flag indicating whether outcome is success or failure.

Returns:

Count of training samples with specified success.

compileTo

public void compileTo(ObjectOutput objOut)
               throws IOException

Compiles this Bernoulli estimator to the specified object output. The corresponding read will produce an instance of BernoulliConstant with the same success probability as the estimate derived from this estimator.

Specified by:

compileTo in interface Compilable

Parameters:

objOut - Object output to which this Bernoulli distribution is written.

Throws:

IOException - If there is an I/O error compiling the object.