org.openimaj.experiment.dataset.sampling

Class GroupedUniformRandomisedSampler<KEY,INSTANCE>

java.lang.Object

org.openimaj.experiment.dataset.sampling.GroupedUniformRandomisedSampler<KEY,INSTANCE>

Type Parameters:

KEY - Type of groups

INSTANCE - Type of instances

All Implemented Interfaces:

Sampler<GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE>>

public class GroupedUniformRandomisedSampler<KEY,INSTANCE>
extends Object
implements Sampler<GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE>>

A uniformly random sampling scheme for grouped datasets. Both sampling with and without replacement are supported. The sampler returns a dataset that selects a predefined fraction of the input data. No attempt is made to ensure that the distribution across groups is maintained (see StratifiedGroupedUniformRandomisedSampler to achieve that).

Author:

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

Constructor Summary

Constructors

Constructor and Description
GroupedUniformRandomisedSampler(double percentage) Construct a GroupedUniformRandomisedSampler with the given percentage of instances to select.
GroupedUniformRandomisedSampler(double percentage, boolean withReplacement) Construct a GroupedUniformRandomisedSampler with the given percentage of instances to select, using with with-replacement or without-replacement sampling.
GroupedUniformRandomisedSampler(int number) Construct a GroupedUniformRandomisedSampler with the given number of instances to select.
GroupedUniformRandomisedSampler(int number, boolean withReplacement) Construct a GroupedUniformRandomisedSampler with the given number of instances to select, using with with-replacement or without-replacement sampling.

Method Summary

Modifier and Type	Method and Description
GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>	sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset) Perform the sampling operation on the given data and return the sampled dataset.
static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>	sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset, double percentage) Sample a dataset with the given percentage of instances to select.
static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>	sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset, double percentage, boolean withReplacement) Sample a dataset with the given percentage of instances to select, using with with-replacement or without-replacement sampling.
static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>	sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset, int number) Sample a dataset with the given number of instances to select.
static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE>	sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset, int number, boolean withReplacement) Sample a dataset with the given number of instances to select, using with with-replacement or without-replacement sampling.

Methods inherited from class java.lang.Object

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

Constructor Detail

GroupedUniformRandomisedSampler

public GroupedUniformRandomisedSampler(double percentage)

Construct a GroupedUniformRandomisedSampler with the given percentage of instances to select. By default, the sampling is without replacement (i.e. an instance can only be selected once).

Parameters:

percentage - percentage of instances to select

GroupedUniformRandomisedSampler

public GroupedUniformRandomisedSampler(double percentage,
                                       boolean withReplacement)

Construct a GroupedUniformRandomisedSampler with the given percentage of instances to select, using with with-replacement or without-replacement sampling.

Parameters:

percentage - percentage of instances to select

withReplacement - should the sampling be performed with replacement (true); or without replacement (false).

GroupedUniformRandomisedSampler

public GroupedUniformRandomisedSampler(int number)

Construct a GroupedUniformRandomisedSampler with the given number of instances to select. By default, the sampling is without replacement (i.e. an instance can only be selected once).

Parameters:

number - number of instances to select

GroupedUniformRandomisedSampler

public GroupedUniformRandomisedSampler(int number,
                                       boolean withReplacement)

Construct a GroupedUniformRandomisedSampler with the given number of instances to select, using with with-replacement or without-replacement sampling.

Parameters:

number - number of instances to select

withReplacement - should the sampling be performed with replacement (true); or without replacement (false).

Method Detail

sample

public GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE> sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset)

Description copied from interface: Sampler

Perform the sampling operation on the given data and return the sampled dataset.

Specified by:

sample in interface Sampler<GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE>>

Parameters:

dataset - the dataset to sample

Returns:

the sampled dataset

sample

public static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE> sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset,
                                                                                       double percentage)

Sample a dataset with the given percentage of instances to select. By default, the sampling is without replacement (i.e. an instance can only be selected once).

Parameters:

dataset - the dataset to sample

percentage - percentage of instances to select

Returns:

the sampled dataset

sample

public static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE> sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset,
                                                                                       double percentage,
                                                                                       boolean withReplacement)

Sample a dataset with the given percentage of instances to select, using with with-replacement or without-replacement sampling.

Parameters:

dataset - the dataset to sample

percentage - percentage of instances to select

withReplacement - should the sampling be performed with replacement (true); or without replacement (false).

Returns:

the sampled dataset

sample

public static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE> sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset,
                                                                                       int number)

Sample a dataset with the given number of instances to select. By default, the sampling is without replacement (i.e. an instance can only be selected once).

Parameters:

dataset - the dataset to sample

number - number of instances to select

Returns:

the sampled dataset

sample

public static <KEY,INSTANCE> GroupedDataset<KEY,ListDataset<INSTANCE>,INSTANCE> sample(GroupedDataset<KEY,? extends ListDataset<INSTANCE>,INSTANCE> dataset,
                                                                                       int number,
                                                                                       boolean withReplacement)

Sample a dataset with the given number of instances to select, using with with-replacement or without-replacement sampling.

Parameters:

dataset - the dataset to sample

number - number of instances to select

withReplacement - should the sampling be performed with replacement (true); or without replacement (false).

Returns:

the sampled dataset