org.tribuo.evaluation

Class CrossValidation<T extends Output<T>,E extends Evaluation<T>>

java.lang.Object

org.tribuo.evaluation.CrossValidation<T,E>

public class CrossValidation<T extends Output<T>,E extends Evaluation<T>>
extends Object

A class that does k-fold cross-validation.

This splits the data into k pieces, tests on one of them and trains on the rest.

It produces a list of Evaluations for each of the test sets.

Constructor Summary

Constructors

Constructor and Description
CrossValidation(Trainer<T> trainer, Dataset<T> data, Evaluator<T,E> evaluator, int k) Builds a k-fold cross-validation loop.
CrossValidation(Trainer<T> trainer, Dataset<T> data, Evaluator<T,E> evaluator, int k, long seed) Builds a k-fold cross-validation loop.

Method Summary

Modifier and Type	Method and Description
List<com.oracle.labs.mlrg.olcut.util.Pair<E,Model<T>>>	evaluate() Performs k fold cross validation, returning the k evaluations.
int	getK() Returns the number of folds.

Methods inherited from class java.lang.Object

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

Constructor Detail

CrossValidation

public CrossValidation(Trainer<T> trainer,
                       Dataset<T> data,
                       Evaluator<T,E> evaluator,
                       int k)

Builds a k-fold cross-validation loop.

Parameters:

trainer - the trainer to use.

data - the dataset to split.

evaluator - the evaluator to use.

k - the number of folds.

CrossValidation

public CrossValidation(Trainer<T> trainer,
                       Dataset<T> data,
                       Evaluator<T,E> evaluator,
                       int k,
                       long seed)

Builds a k-fold cross-validation loop.

Parameters:

trainer - the trainer to use.

data - the dataset to split.

evaluator - the evaluator to use.

k - the number of folds.

seed - The RNG seed.

Method Detail

getK

public int getK()

Returns the number of folds.

Returns:

The number of folds.

evaluate

public List<com.oracle.labs.mlrg.olcut.util.Pair<E,Model<T>>> evaluate()

Performs k fold cross validation, returning the k evaluations.

Returns:

The k evaluators one per fold.