ai.libs.jaicore.ml

Class WekaUtil

java.lang.Object

ai.libs.jaicore.ml.WekaUtil

public class WekaUtil
extends java.lang.Object

Constructor Summary

Constructors

Constructor and Description
WekaUtil()

Method Summary

Modifier and Type	Method and Description
static weka.classifiers.Classifier	cloneClassifier(weka.classifiers.Classifier c)
static weka.core.Instance	fromJAICoreInstance(Instance instance)
static weka.core.Instance	fromJAICoreInstance(LabeledInstance<java.lang.String> instance)
static weka.core.Instances	fromJAICoreInstances(Instances instances)
static weka.core.Instances	fromJAICoreInstances(LabeledInstances<java.lang.String> labeledInstances)
static <L> weka.core.Instances	fromJAICoreInstances(WekaCompatibleInstancesImpl instances)
static java.util.Collection<java.util.List<java.lang.String>>	getAdmissibleSearcherEvaluatorCombinationsForAttributeSelection() Determines all attribute selection variants (search/evaluator combinations with default parametrization)
static java.util.Collection<java.lang.Integer>[]	getArbitrarySplit(weka.core.Instances data, java.util.Random rand, double... portions)
static java.util.List<weka.core.Attribute>	getAttributes(weka.core.Instance inst)
static java.util.List<weka.core.Attribute>	getAttributes(weka.core.Instances inst, boolean includeClassAttribute)
static java.util.Collection<java.lang.String>	getBasicLearners()
static java.util.Collection<java.lang.String>	getBinaryClassifiers()
static java.util.Collection<java.lang.String>	getClassesActuallyContainedInDataset(weka.core.Instances data)
static double[]	getClassesAsArray(weka.core.Instances inst)
static java.util.List<java.lang.Double>	getClassesAsList(weka.core.Instances inst)
static java.util.List<java.lang.String>	getClassesDeclaredInDataset(weka.core.Instances data)
static java.lang.String	getClassifierDescriptor(weka.classifiers.Classifier c)
static java.lang.String	getClassName(weka.core.Instance instance)
static java.util.List<java.lang.String>	getClassNames(weka.core.Instance instance)
static java.util.Map<java.lang.String,java.lang.Integer>	getClassNameToIDMap(weka.core.Instance instance)
static java.util.List<java.io.File>	getDatasetsInFolder(java.io.File folder)
static weka.core.Instances	getEmptyDatasetForJAICoreInstance(Instance instance)
static weka.core.Instances	getEmptySetOfInstancesWithRefactoredClass(weka.core.Instances instances)
static weka.core.Instances	getEmptySetOfInstancesWithRefactoredClass(weka.core.Instances instances, java.util.List<java.lang.String> classes)
static java.util.Collection<java.lang.String>	getFeatureEvaluators()
static int[]	getIndicesOfContainedInstances(weka.core.Instances dataset, weka.core.Instances subset) Compute indices of instances of the original data set that are contained in the given subset.
static int[]	getIndicesOfSubset(weka.core.Instances data, weka.core.Instances subset)
static weka.core.Instances	getInstancesOfClass(weka.core.Instances data, java.util.Collection<java.lang.String> classNames)
static weka.core.Instances	getInstancesOfClass(weka.core.Instances data, java.lang.String className)
static java.util.Map<java.lang.String,weka.core.Instances>	getInstancesPerClass(weka.core.Instances data)
static int	getIntValOfClassName(weka.core.Instance instance, java.lang.String className)
static java.util.Collection<java.lang.String>	getMetaLearners()
static java.util.Collection<java.lang.String>	getNativeMultiClassClassifiers()
static weka.core.Attribute	getNewClassAttribute(weka.core.Attribute attribute)
static weka.core.Attribute	getNewClassAttribute(weka.core.Attribute attribute, java.util.List<java.lang.String> classes)
static int	getNumberOfInstancesFromClass(weka.core.Instances data, java.util.Collection<java.lang.String> cs)
static int	getNumberOfInstancesFromClass(weka.core.Instances data, java.lang.String c)
static java.util.Map<java.lang.String,java.lang.Integer>	getNumberOfInstancesPerClass(weka.core.Instances data)
static java.util.Collection<weka.core.Option>	getOptionsOfWekaAlgorithm(java.lang.Object o)
static java.util.Collection<java.lang.String>	getPossibleClassValues(weka.core.Instance instance)
static weka.core.Instance	getRefactoredInstance(weka.core.Instance instance)
static weka.core.Instance	getRefactoredInstance(weka.core.Instance instance, java.util.List<java.lang.String> classes)
static weka.core.Instances	getRefactoredInstances(weka.core.Instances data, java.util.Map<java.lang.String,java.lang.String> classMap)
static double	getRelativeNumberOfInstancesFromClass(weka.core.Instances data, java.util.Collection<java.lang.String> cs)
static double	getRelativeNumberOfInstancesFromClass(weka.core.Instances data, java.lang.String c)
static java.util.List<weka.core.Attribute>	getReplacedAttributeList(java.util.List<weka.core.Attribute> attributes, weka.core.Attribute classAttribute)
static java.util.Collection<java.lang.String>	getSearchers()
static java.util.List<weka.core.Instances>	getStratifiedSplit(weka.core.Instances data, long seed, double... portions)
static java.util.List<ReproducibleInstances>	getStratifiedSplit(ReproducibleInstances data, long seed, double... portions) Creates a StratifiedSplit for a given ReproducibleInstances Object.
static java.util.List<ReproducibleInstances>	getStratifiedSplit(ReproducibleInstances data, java.util.Random rand, double... portions) Creates a stratified split for a given ReproducibleInstances Object.
static java.util.Collection<java.lang.Integer>[]	getStratifiedSplitIndices(weka.core.Instances data, java.util.Random rand, double... pPortions)
static java.util.List<java.util.List<java.lang.Integer>>	getStratifiedSplitIndicesAsList(weka.core.Instances data, java.util.Random rand, double... portions)
static boolean	hasOnlyNumericAttributes(weka.core.Instances instances)
static java.lang.String	instancesToJsonString(weka.core.Instances data)
static boolean	isValidPreprocessorCombination(java.lang.String searcher, java.lang.String evaluator)
static weka.core.Instances	jsonStringToInstances(java.lang.String json)
static weka.core.Instances	mergeClassesOfInstances(weka.core.Instances data, java.util.Collection<java.lang.String> cluster1, java.util.Collection<java.lang.String> cluster2)
static weka.core.Instances	mergeClassesOfInstances(weka.core.Instances data, java.util.List<java.util.Set<java.lang.String>> instancesCluster)
static boolean	needsBinarization(weka.core.Instances wekaInstances, boolean ignoreClassAttribute) Returns true if there is at least one nominal attribute in the given dataset that has more than 2 values.
static void	printClassSplitAssignments(java.util.List<weka.core.Instances> split)
static java.lang.String	printNestedWekaClassifier(weka.classifiers.Classifier c)
static java.util.List<weka.core.Instances>	realizeSplit(weka.core.Instances data, java.util.Collection<java.lang.Integer>[] split)
static java.util.List<weka.core.Instances>	realizeSplit(weka.core.Instances data, java.util.List<java.util.List<java.lang.Integer>> split)
static java.util.List<weka.core.Instances>	realizeSplitAsCopiedInstances(weka.core.Instances data, java.util.Collection<java.lang.Integer>[] split)
static java.util.List<weka.core.Instances>	realizeSplitAsCopiedInstances(weka.core.Instances data, java.util.List<java.util.List<java.lang.Integer>> split)
static java.util.List<weka.core.Instances>	realizeSplitAsSubInstances(weka.core.Instances data, java.util.Collection<java.lang.Integer>[] split)
static weka.core.Instance	removeClassAttribute(weka.core.Instance inst)
static weka.core.Instances	removeClassAttribute(weka.core.Instances data)
static com.fasterxml.jackson.databind.node.ArrayNode	splitToJsonArray(java.util.Collection<java.lang.Integer>[] splitDecision)
static Instance	toJAICoreInstance(weka.core.Instance wekaInst)
static Instances	toJAICoreInstances(weka.core.Instances wekaInstances)
static LabeledInstance<java.lang.String>	toJAICoreLabeledInstance(weka.core.Instance wekaInst)
static WekaCompatibleInstancesImpl	toJAICoreLabeledInstances(weka.core.Instances wekaInstances)
static weka.core.Instance	useFilterOnSingleInstance(weka.core.Instance instance, weka.filters.Filter filter)

Methods inherited from class java.lang.Object

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

Constructor Detail

WekaUtil

public WekaUtil()

Method Detail

getBasicLearners

public static java.util.Collection<java.lang.String> getBasicLearners()

getNativeMultiClassClassifiers

public static java.util.Collection<java.lang.String> getNativeMultiClassClassifiers()

getBinaryClassifiers

public static java.util.Collection<java.lang.String> getBinaryClassifiers()

getFeatureEvaluators

public static java.util.Collection<java.lang.String> getFeatureEvaluators()

getSearchers

public static java.util.Collection<java.lang.String> getSearchers()

getMetaLearners

public static java.util.Collection<java.lang.String> getMetaLearners()

isValidPreprocessorCombination

public static boolean isValidPreprocessorCombination(java.lang.String searcher,
                                                     java.lang.String evaluator)

getAdmissibleSearcherEvaluatorCombinationsForAttributeSelection

public static java.util.Collection<java.util.List<java.lang.String>> getAdmissibleSearcherEvaluatorCombinationsForAttributeSelection()

Determines all attribute selection variants (search/evaluator combinations with default parametrization)

Returns:

fromJAICoreInstances

public static <L> weka.core.Instances fromJAICoreInstances(WekaCompatibleInstancesImpl instances)

fromJAICoreInstances

public static weka.core.Instances fromJAICoreInstances(Instances instances)

getEmptyDatasetForJAICoreInstance

public static weka.core.Instances getEmptyDatasetForJAICoreInstance(Instance instance)

fromJAICoreInstance

public static weka.core.Instance fromJAICoreInstance(Instance instance)

fromJAICoreInstance

public static weka.core.Instance fromJAICoreInstance(LabeledInstance<java.lang.String> instance)

fromJAICoreInstances

public static weka.core.Instances fromJAICoreInstances(LabeledInstances<java.lang.String> labeledInstances)

toJAICoreLabeledInstances

public static WekaCompatibleInstancesImpl toJAICoreLabeledInstances(weka.core.Instances wekaInstances)

needsBinarization

public static boolean needsBinarization(weka.core.Instances wekaInstances,
                                        boolean ignoreClassAttribute)

Returns true if there is at least one nominal attribute in the given dataset that has more than 2 values.

Parameters:

wekaInstances - dataset that is checked

ignoreClassAttribute - if true class attribute is ignored.

toJAICoreLabeledInstance

public static LabeledInstance<java.lang.String> toJAICoreLabeledInstance(weka.core.Instance wekaInst)

toJAICoreInstances

public static Instances toJAICoreInstances(weka.core.Instances wekaInstances)

getPossibleClassValues

public static java.util.Collection<java.lang.String> getPossibleClassValues(weka.core.Instance instance)

toJAICoreInstance

public static Instance toJAICoreInstance(weka.core.Instance wekaInst)

getClassifierDescriptor

public static java.lang.String getClassifierDescriptor(weka.classifiers.Classifier c)

getOptionsOfWekaAlgorithm

public static java.util.Collection<weka.core.Option> getOptionsOfWekaAlgorithm(java.lang.Object o)

getClassNames

public static java.util.List<java.lang.String> getClassNames(weka.core.Instance instance)

getClassNameToIDMap

public static java.util.Map<java.lang.String,java.lang.Integer> getClassNameToIDMap(weka.core.Instance instance)

getIntValOfClassName

public static int getIntValOfClassName(weka.core.Instance instance,
                                       java.lang.String className)

printClassSplitAssignments

public static void printClassSplitAssignments(java.util.List<weka.core.Instances> split)

getInstancesOfClass

public static weka.core.Instances getInstancesOfClass(weka.core.Instances data,
                                                      java.util.Collection<java.lang.String> classNames)

getInstancesOfClass

public static weka.core.Instances getInstancesOfClass(weka.core.Instances data,
                                                      java.lang.String className)

getClassName

public static java.lang.String getClassName(weka.core.Instance instance)

getInstancesPerClass

public static java.util.Map<java.lang.String,weka.core.Instances> getInstancesPerClass(weka.core.Instances data)

getNumberOfInstancesPerClass

public static java.util.Map<java.lang.String,java.lang.Integer> getNumberOfInstancesPerClass(weka.core.Instances data)

getNumberOfInstancesFromClass

public static int getNumberOfInstancesFromClass(weka.core.Instances data,
                                                java.lang.String c)

getNumberOfInstancesFromClass

public static int getNumberOfInstancesFromClass(weka.core.Instances data,
                                                java.util.Collection<java.lang.String> cs)

getRelativeNumberOfInstancesFromClass

public static double getRelativeNumberOfInstancesFromClass(weka.core.Instances data,
                                                           java.lang.String c)

getRelativeNumberOfInstancesFromClass

public static double getRelativeNumberOfInstancesFromClass(weka.core.Instances data,
                                                           java.util.Collection<java.lang.String> cs)

getArbitrarySplit

public static java.util.Collection<java.lang.Integer>[] getArbitrarySplit(weka.core.Instances data,
                                                                          java.util.Random rand,
                                                                          double... portions)

realizeSplit

public static java.util.List<weka.core.Instances> realizeSplit(weka.core.Instances data,
                                                               java.util.Collection<java.lang.Integer>[] split)

realizeSplit

public static java.util.List<weka.core.Instances> realizeSplit(weka.core.Instances data,
                                                               java.util.List<java.util.List<java.lang.Integer>> split)

realizeSplitAsCopiedInstances

public static java.util.List<weka.core.Instances> realizeSplitAsCopiedInstances(weka.core.Instances data,
                                                                                java.util.List<java.util.List<java.lang.Integer>> split)

realizeSplitAsCopiedInstances

public static java.util.List<weka.core.Instances> realizeSplitAsCopiedInstances(weka.core.Instances data,
                                                                                java.util.Collection<java.lang.Integer>[] split)

realizeSplitAsSubInstances

public static java.util.List<weka.core.Instances> realizeSplitAsSubInstances(weka.core.Instances data,
                                                                             java.util.Collection<java.lang.Integer>[] split)

getStratifiedSplitIndices

public static java.util.Collection<java.lang.Integer>[] getStratifiedSplitIndices(weka.core.Instances data,
                                                                                  java.util.Random rand,
                                                                                  double... pPortions)

getStratifiedSplitIndicesAsList

public static java.util.List<java.util.List<java.lang.Integer>> getStratifiedSplitIndicesAsList(weka.core.Instances data,
                                                                                                java.util.Random rand,
                                                                                                double... portions)

splitToJsonArray

public static com.fasterxml.jackson.databind.node.ArrayNode splitToJsonArray(java.util.Collection<java.lang.Integer>[] splitDecision)

getStratifiedSplit

public static java.util.List<weka.core.Instances> getStratifiedSplit(weka.core.Instances data,
                                                                     long seed,
                                                                     double... portions)

getStratifiedSplit

public static java.util.List<ReproducibleInstances> getStratifiedSplit(ReproducibleInstances data,
                                                                       java.util.Random rand,
                                                                       double... portions)

Creates a stratified split for a given ReproducibleInstances Object. The history will be updated to track the split.

Parameters:

data - - Input data

rand - - random used to get a seed, which can be used and saved

portions - - ratios to split

Returns:

a list of ReproducibleInstances. For each of them the history will be updated to track the split

getStratifiedSplit

public static java.util.List<ReproducibleInstances> getStratifiedSplit(ReproducibleInstances data,
                                                                       long seed,
                                                                       double... portions)

Creates a StratifiedSplit for a given ReproducibleInstances Object. THe History will be updated to track the split.

Parameters:

data - - Input data

seed - - random seed

portions - - ratios to split

Returns:

a List of ReproducibleInstances. For each of them the history will be updated to track the split

getDatasetsInFolder

public static java.util.List<java.io.File> getDatasetsInFolder(java.io.File folder)
                                                        throws java.io.IOException

Throws:

java.io.IOException

getRefactoredInstances

public static weka.core.Instances getRefactoredInstances(weka.core.Instances data,
                                                         java.util.Map<java.lang.String,java.lang.String> classMap)

getRefactoredInstance

public static weka.core.Instance getRefactoredInstance(weka.core.Instance instance)

getRefactoredInstance

public static weka.core.Instance getRefactoredInstance(weka.core.Instance instance,
                                                       java.util.List<java.lang.String> classes)

getEmptySetOfInstancesWithRefactoredClass

public static weka.core.Instances getEmptySetOfInstancesWithRefactoredClass(weka.core.Instances instances)

getEmptySetOfInstancesWithRefactoredClass

public static weka.core.Instances getEmptySetOfInstancesWithRefactoredClass(weka.core.Instances instances,
                                                                            java.util.List<java.lang.String> classes)

getAttributes

public static java.util.List<weka.core.Attribute> getAttributes(weka.core.Instances inst,
                                                                boolean includeClassAttribute)

getAttributes

public static java.util.List<weka.core.Attribute> getAttributes(weka.core.Instance inst)

hasOnlyNumericAttributes

public static boolean hasOnlyNumericAttributes(weka.core.Instances instances)

getNewClassAttribute

public static weka.core.Attribute getNewClassAttribute(weka.core.Attribute attribute)

getNewClassAttribute

public static weka.core.Attribute getNewClassAttribute(weka.core.Attribute attribute,
                                                       java.util.List<java.lang.String> classes)

getReplacedAttributeList

public static java.util.List<weka.core.Attribute> getReplacedAttributeList(java.util.List<weka.core.Attribute> attributes,
                                                                           weka.core.Attribute classAttribute)

mergeClassesOfInstances

public static weka.core.Instances mergeClassesOfInstances(weka.core.Instances data,
                                                          java.util.Collection<java.lang.String> cluster1,
                                                          java.util.Collection<java.lang.String> cluster2)

mergeClassesOfInstances

public static weka.core.Instances mergeClassesOfInstances(weka.core.Instances data,
                                                          java.util.List<java.util.Set<java.lang.String>> instancesCluster)

getClassesDeclaredInDataset

public static java.util.List<java.lang.String> getClassesDeclaredInDataset(weka.core.Instances data)

getClassesActuallyContainedInDataset

public static java.util.Collection<java.lang.String> getClassesActuallyContainedInDataset(weka.core.Instances data)

getClassesAsArray

public static double[] getClassesAsArray(weka.core.Instances inst)

getClassesAsList

public static java.util.List<java.lang.Double> getClassesAsList(weka.core.Instances inst)

instancesToJsonString

public static java.lang.String instancesToJsonString(weka.core.Instances data)

jsonStringToInstances

public static weka.core.Instances jsonStringToInstances(java.lang.String json)

getIndicesOfContainedInstances

public static int[] getIndicesOfContainedInstances(weka.core.Instances dataset,
                                                   weka.core.Instances subset)

Compute indices of instances of the original data set that are contained in the given subset. This does only work for data sets that contain an instance at most once!

Parameters:

dataset -

subset -

Returns:

useFilterOnSingleInstance

public static weka.core.Instance useFilterOnSingleInstance(weka.core.Instance instance,
                                                           weka.filters.Filter filter)
                                                    throws java.lang.Exception

Throws:

java.lang.Exception

removeClassAttribute

public static weka.core.Instances removeClassAttribute(weka.core.Instances data)
                                                throws java.lang.Exception

Throws:

java.lang.Exception

removeClassAttribute

public static weka.core.Instance removeClassAttribute(weka.core.Instance inst)
                                               throws java.lang.Exception

Throws:

java.lang.Exception

cloneClassifier

public static weka.classifiers.Classifier cloneClassifier(weka.classifiers.Classifier c)
                                                   throws java.lang.Exception

Throws:

java.lang.Exception

getIndicesOfSubset

public static int[] getIndicesOfSubset(weka.core.Instances data,
                                       weka.core.Instances subset)

printNestedWekaClassifier

public static java.lang.String printNestedWekaClassifier(weka.classifiers.Classifier c)