ai.libs.jaicore.ml.weka.dataset.splitter

Interface IMultilabelCrossValidation

All Known Implementing Classes:

RandomMultilabelCrossValidation

public interface IMultilabelCrossValidation

Represents an algorithm that realizes a split of a given multilabel instances in folds, given a seed, custom information about the split represented as a string, and the fold that is left out for testing.

Method Summary

Modifier and Type	Method and Description
default java.lang.String	generateSplittingInfo(double... portions) Generate a string representation that represents only the split info part of the split string.
default java.lang.String	generateSplittingString(double... portions) Generate a String that represents a split of a data set into portions from the given portions sizes (must add up to <1).
java.lang.String	getName() Get the name of the implementing multilabel cross validation technique.
java.lang.String	getSplitSeparator() Get the separator used to separate single portions of a split in a given splitInfo.
weka.core.Instances	getTestSplit(weka.core.Instances data, int seed, int fold, java.lang.String splitInfo) Gets a test split from the given data based on the seed.
weka.core.Instances	getTrainSplit(weka.core.Instances data, int seed, int fold, java.lang.String splitInfo) Gets a train split from the given data based on the seed.

Method Detail

getTestSplit

weka.core.Instances getTestSplit(weka.core.Instances data,
                                 int seed,
                                 int fold,
                                 java.lang.String splitInfo)

Gets a test split from the given data based on the seed. The given fold is the test fold.

Parameters:

data - The data from which to derive the split

seed - The seed possibly used by the implementation class to derive the split

fold - The number of the fold which is the testing fold

splitInfo - Information about the split for the class executing the split (e.g. portion sizes of folds)

Returns:

A test Instances derived from the given instances

getTrainSplit

weka.core.Instances getTrainSplit(weka.core.Instances data,
                                  int seed,
                                  int fold,
                                  java.lang.String splitInfo)

Gets a train split from the given data based on the seed. The given fold is the test fold, which is the left out fold for this case.

Parameters:

data - The data from which to derive the split

seed - The seed possibly used by the implementation class to derive the split

fold - The number of the fold which is the testing fold (left out fold)

splitInfo - Information about the split for the class executing the split (e.g. portion sized of folds)

Returns:

A train Instances derived from the given instances

generateSplittingString

default java.lang.String generateSplittingString(double... portions)

Generate a String that represents a split of a data set into portions from the given portions sizes (must add up to <1).

Parameters:

portions - The portions sizes of the split

Returns:

The String representation of the split

generateSplittingInfo

default java.lang.String generateSplittingInfo(double... portions)

Generate a string representation that represents only the split info part of the split string.

Parameters:

portions - portions sized of the split

Returns:

String representation of the splitInfo part of the split

getName

java.lang.String getName()

Get the name of the implementing multilabel cross validation technique.

Returns:

The name of the technique

getSplitSeparator

java.lang.String getSplitSeparator()

Get the separator used to separate single portions of a split in a given splitInfo. E.g. if the splitInfo represents data portions, it could be 0.7/0.3 with "/" being the split separator.

Returns:

The slit separator used by the specific implementation of this interface