de.lmu.ifi.dbs.elki.evaluation.classification.holdout

Interface Holdout

All Known Implementing Classes:

AbstractHoldout, DisjointCrossValidation, LeaveOneOut, RandomizedCrossValidation, RandomizedHoldout, StratifiedCrossValidation

public interface Holdout

A holdout procedure is to provide a range of partitions of a database to pairs of training and test data sets.

Author:

Erich Schubert

Method Summary

Modifier and Type	Method and Description
ArrayList<ClassLabel>	getLabels() Get the sorted class labels present in this data set.
void	initialize(MultipleObjectsBundle bundle) Initialize the holdout procedure for a data set.
TrainingAndTestSet	nextPartitioning() Get the next partitioning of the given holdout.
int	numberOfPartitions() How many partitions to test.

Method Detail

initialize

void initialize(MultipleObjectsBundle bundle)

Initialize the holdout procedure for a data set.

Parameters:

bundle - Data set bundle

nextPartitioning

TrainingAndTestSet nextPartitioning()

Get the next partitioning of the given holdout.

Returns:

Next partitioning of the data set

getLabels

ArrayList<ClassLabel> getLabels()

Get the sorted class labels present in this data set. For indexing into assignment arrays.

Returns:

Class labels

numberOfPartitions

int numberOfPartitions()

How many partitions to test.

Returns:

Number of partitions.