org.openimaj.experiment.validation

Class StratifiedGroupedRandomisedPercentageHoldOut<KEY,INSTANCE>

java.lang.Object

org.openimaj.experiment.validation.DefaultValidationData<GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>>

org.openimaj.experiment.validation.StratifiedGroupedRandomisedPercentageHoldOut<KEY,INSTANCE>

Type Parameters:

KEY - Type of groups

INSTANCE - Type of the instances in the dataset

All Implemented Interfaces:

TrainSplitProvider<GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>>, ValidateSplitProvider<GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>>, ValidationData<GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>>

public class StratifiedGroupedRandomisedPercentageHoldOut<KEY,INSTANCE>
extends DefaultValidationData<GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>>

Stratified Hold-Out validation for grouped data that selects a percentage of the original data to use for training, and the remainder to use for validation. The splitting of data is performed per group to ensure that the relative group sizes remain (approximately) constant.

Author:

Jonathon Hare (jsh2@ecs.soton.ac.uk)

Field Summary

Fields inherited from class org.openimaj.experiment.validation.DefaultValidationData

training, validation

Constructor Summary

Constructors

Constructor and Description
StratifiedGroupedRandomisedPercentageHoldOut(double percentageTraining, GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE> dataset) Construct with the given dataset and percentage of training data (0..1).

Method Summary

Methods inherited from class org.openimaj.experiment.validation.DefaultValidationData

getTrainingDataset, getValidationDataset

Methods inherited from class java.lang.Object

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

Constructor Detail

StratifiedGroupedRandomisedPercentageHoldOut

public StratifiedGroupedRandomisedPercentageHoldOut(double percentageTraining,
                                            GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE> dataset)

Construct with the given dataset and percentage of training data (0..1).

Parameters:

percentageTraining - percentage of the dataset to use for training

dataset - the dataset