org.cogroo.tools.featurizer

Class FeaturizerCrossValidator

java.lang.Object

org.cogroo.tools.featurizer.FeaturizerCrossValidator

public class FeaturizerCrossValidator
extends Object

Constructor Summary

Constructors

Constructor and Description
FeaturizerCrossValidator(String languageCode, opennlp.tools.util.TrainingParameters params, FeatureDictionary dict, String cgFlags, String factoryClass, FeaturizerEvaluationMonitor... listeners)

Method Summary

Modifier and Type	Method and Description
void	evaluate(opennlp.tools.util.ObjectStream<FeatureSample> samples, int nFolds) Starts the evaluation.
double	getWordAccuracy() Retrieves the accuracy for all iterations.
long	getWordCount() Retrieves the number of words which where validated over all iterations.

Methods inherited from class java.lang.Object

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

Constructor Detail

FeaturizerCrossValidator

public FeaturizerCrossValidator(String languageCode,
                                opennlp.tools.util.TrainingParameters params,
                                FeatureDictionary dict,
                                String cgFlags,
                                String factoryClass,
                                FeaturizerEvaluationMonitor... listeners)

Method Detail

evaluate

public void evaluate(opennlp.tools.util.ObjectStream<FeatureSample> samples,
                     int nFolds)
              throws IOException,
                     opennlp.tools.util.InvalidFormatException,
                     IOException

Starts the evaluation.

Parameters:

samples - the data to train and test

nFolds - number of folds

Throws:

IOException

opennlp.tools.util.InvalidFormatException

getWordAccuracy

public double getWordAccuracy()

Retrieves the accuracy for all iterations.

Returns:

the word accuracy

getWordCount

public long getWordCount()

Retrieves the number of words which where validated over all iterations. The result is the amount of folds multiplied by the total number of words.

Returns:

the word count