Package com.tibco.patterns.learn.api.hint

Class DatasetStats

java.lang.Object

com.tibco.patterns.learn.api.hint.DatasetStats

public final class DatasetStats
extends java.lang.Object

Stores statistics about existing pairs in a dataset, grouped by subset: the number of pairs with each label and the total number of labeled pairs. Used to determine whether to add another pair to the dataset.

Constructor Summary

Constructors

Constructor	Description
DatasetStats(FeatureQuery featureQuery, RecPairMap recPairMap, DataPartition partition)	Calculates and stores all dataset statistics for the given dataset.

Method Summary

Modifier and Type	Method	Description
static SubsetFamily	calcDatasetSubsets(FeatureQuery fq, RecPairMap recPairMap, DataPartition partition, boolean boolLabelsOnly)	Finds all subsets present in the specified dataset.
SubsetFamily	getFamilyAllSubsets()
DataPartition	getPartition()
java.lang.String	toString()
void	update(RecPair newRecPair)	Updates statistics with the info about the pair that has been added to this dataset.

Methods inherited from class java.lang.Object

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

Constructor Detail

DatasetStats

public DatasetStats(FeatureQuery featureQuery,
                    RecPairMap recPairMap,
                    DataPartition partition)

Calculates and stores all dataset statistics for the given dataset.

Parameters:

recPairMap - - stores all pairs in the dataset.

featureQuery - - stores all features.

Method Detail

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getPartition

public DataPartition getPartition()

Returns:

the data partition that this object was created for.

getFamilyAllSubsets

public SubsetFamily getFamilyAllSubsets()

Returns:

the internal subset family that stores all subsets in the dataset.

calcDatasetSubsets

public static SubsetFamily calcDatasetSubsets(FeatureQuery fq,
                                              RecPairMap recPairMap,
                                              DataPartition partition,
                                              boolean boolLabelsOnly)

Finds all subsets present in the specified dataset. Only pairs that have feature values are counted as belonging to a subset.

Parameters:

boolLabelsOnly - - if true, only pairs with bool labels are counted.

Returns:

a new subset family that contains these subsets.

update

public void update(RecPair newRecPair)

Updates statistics with the info about the pair that has been added to this dataset. Must be called after adding each pair to keep statistics current.

Parameters:

newRecPair - - the pair that has just been added to this dataset.

Throws:

java.lang.IllegalArgumentException - if recPair does not contain a boolean label or feature scores.