org.apache.ignite.ml.structures

Class LabeledVectorSet<Row extends LabeledVector>

java.lang.Object

org.apache.ignite.ml.structures.Dataset<Row>

org.apache.ignite.ml.structures.LabeledVectorSet<Row>

All Implemented Interfaces:

Externalizable, Serializable, AutoCloseable

public class LabeledVectorSet<Row extends LabeledVector>
extends Dataset<Row>
implements AutoCloseable

The set of labeled vectors used in local partition calculations.

See Also:

Serialized Form

Field Summary

Fields inherited from class org.apache.ignite.ml.structures.Dataset

colSize, data, isDistributed, meta, rowSize

Constructor Summary

Constructors

Constructor and Description
LabeledVectorSet() Default constructor (required by Externalizable).
LabeledVectorSet(double[][] mtx, double[] lbs) Creates new local Labeled Dataset by matrix and vector of labels.
LabeledVectorSet(double[][] mtx, double[] lbs, String[] featureNames, boolean isDistributed) Creates new Labeled Dataset by matrix and vector of labels.
LabeledVectorSet(int rowSize, int colSize) Creates new local Labeled Dataset and initialized with empty data structure.
LabeledVectorSet(int rowSize, int colSize, boolean isDistributed) Creates new Labeled Dataset and initialized with empty data structure.
LabeledVectorSet(int rowSize, int colSize, String[] featureNames, boolean isDistributed) Creates new Labeled Dataset and initialized with empty data structure.
LabeledVectorSet(Row[] data) Creates new Labeled Dataset by given data.
LabeledVectorSet(Row[] data, int colSize) Creates new Labeled Dataset by given data.

Method Summary

Modifier and Type	Method and Description
void	close() Closes LabeledDataset.
LabeledVectorSet	copy() Makes copy with new Label objects and old features and Metadata objects.
static Vector	emptyVector(int size, boolean isDistributed)
double	label(int idx) Returns label if label is attached or null if label is missed.
double[]	labels() Returns new copy of labels of all labeled vectors NOTE: This method is useful for copying labels from test dataset.
void	setLabel(int idx, double lb) Fill the label with given value.

Methods inherited from class org.apache.ignite.ml.structures.Dataset

colSize, convertStringNamesToFeatureMetadata, data, equals, features, generateFeatureNames, getFeatureName, getRow, hashCode, isDistributed, meta, readExternal, rowSize, setData, setDistributed, setMeta, writeExternal

Methods inherited from class java.lang.Object

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

Constructor Detail

LabeledVectorSet

public LabeledVectorSet()

Default constructor (required by Externalizable).

LabeledVectorSet

public LabeledVectorSet(int rowSize,
                        int colSize,
                        boolean isDistributed)

Creates new Labeled Dataset and initialized with empty data structure.

Parameters:

rowSize - Amount of instances. Should be > 0.

colSize - Amount of attributes. Should be > 0.

isDistributed - Use distributed data structures to keep data.

LabeledVectorSet

public LabeledVectorSet(int rowSize,
                        int colSize)

Creates new local Labeled Dataset and initialized with empty data structure.

Parameters:

rowSize - Amount of instances. Should be > 0.

colSize - Amount of attributes. Should be > 0.

LabeledVectorSet

public LabeledVectorSet(int rowSize,
                        int colSize,
                        String[] featureNames,
                        boolean isDistributed)

Creates new Labeled Dataset and initialized with empty data structure.

Parameters:

rowSize - Amount of instances. Should be > 0.

colSize - Amount of attributes. Should be > 0

featureNames - Column names.

isDistributed - Use distributed data structures to keep data.

LabeledVectorSet

public LabeledVectorSet(Row[] data)

Creates new Labeled Dataset by given data.

Parameters:

data - Should be initialized with one vector at least.

LabeledVectorSet

public LabeledVectorSet(Row[] data,
                        int colSize)

Creates new Labeled Dataset by given data.

Parameters:

data - Should be initialized with one vector at least.

colSize - Amount of observed attributes in each vector.

LabeledVectorSet

public LabeledVectorSet(double[][] mtx,
                        double[] lbs)

Creates new local Labeled Dataset by matrix and vector of labels.

Parameters:

mtx - Given matrix with rows as observations.

lbs - Labels of observations.

LabeledVectorSet

public LabeledVectorSet(double[][] mtx,
                        double[] lbs,
                        String[] featureNames,
                        boolean isDistributed)

Creates new Labeled Dataset by matrix and vector of labels.

Parameters:

mtx - Given matrix with rows as observations.

lbs - Labels of observations.

featureNames - Column names.

isDistributed - Use distributed data structures to keep data.

Method Detail

label

public double label(int idx)

Returns label if label is attached or null if label is missed.

Parameters:

idx - Index of observation.

Returns:

Label.

labels

public double[] labels()

Returns new copy of labels of all labeled vectors NOTE: This method is useful for copying labels from test dataset.

Returns:

Copy of labels.

setLabel

public void setLabel(int idx,
                     double lb)

Fill the label with given value.

Parameters:

idx - Index of observation.

lb - The given label.

emptyVector

public static Vector emptyVector(int size,
                                 boolean isDistributed)

copy

public LabeledVectorSet copy()

Makes copy with new Label objects and old features and Metadata objects.

close

public void close()
           throws Exception

Closes LabeledDataset.

Specified by:

close in interface AutoCloseable

Throws:

Exception