ml.dmlc.xgboost4j.java

Class DMatrix

java.lang.Object

ml.dmlc.xgboost4j.java.DMatrix

public class DMatrix
extends Object

DMatrix for xgboost.

Author:

hzx

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
class	DMatrix.QuantileCut A class to hold the quantile information
static class	DMatrix.SparseType sparse matrix type (CSR or CSC)

Field Summary

Fields

Modifier and Type	Field and Description
protected long	handle

Constructor Summary

Constructors

Modifier	Constructor and Description
	DMatrix(BigDenseMatrix matrix) create DMatrix from a BigDenseMatrix
	DMatrix(BigDenseMatrix matrix, float missing) create DMatrix from dense matrix
	DMatrix(ColumnBatch columnBatch, float missing, int nthread) Create the normal DMatrix from column array interface
	DMatrix(float[] data, int nrow, int ncol) Deprecated. Please specify the missing value explicitly using DMatrix(float[], int, int, float)
	DMatrix(float[] data, int nrow, int ncol, float missing) create DMatrix from dense matrix
	DMatrix(Iterator<LabeledPoint> iter, String cacheInfo) Create DMatrix from iterator.
	DMatrix(Iterator<LabeledPoint> iter, String cacheInfo, float missing) Create DMatrix from iterator.
protected	DMatrix(long handle) used for DMatrix slice
	DMatrix(long[] headers, int[] indices, float[] data, DMatrix.SparseType st) Deprecated.
	DMatrix(long[] headers, int[] indices, float[] data, DMatrix.SparseType st, int shapeParam) Create DMatrix from Sparse matrix in CSR/CSC format.
	DMatrix(long[] headers, int[] indices, float[] data, DMatrix.SparseType st, int shapeParam, float missing, int nthread)
	DMatrix(String dataPath) Create DMatrix by loading libsvm file from dataPath

Method Summary

Modifier and Type	Method and Description
void	dispose()
protected void	finalize()
float[]	getBaseMargin() Get base margin of the DMatrix.
String[]	getFeatureNames() Get feature names
String[]	getFeatureTypes() Get feature types
int[]	getGroup() Get group sizes of DMatrix
long	getHandle() Get the handle
float[]	getLabel() get label values
DMatrix.QuantileCut	getQuantileCut() Get the Quantile Cut.
float[]	getWeight() get weight of the DMatrix
long	nonMissingNum() Get the number of non-missing values of DMatrix.
long	rowNum() get the row number of DMatrix
void	saveBinary(String filePath) save DMatrix to filePath
void	setBaseMargin(Column column) Set base margin of DMatrix from array interface
void	setBaseMargin(float[] baseMargin) Set base margin (initial prediction).
void	setBaseMargin(float[][] baseMargin) Set base margin (initial prediction).
void	setFeatureNames(String[] values) Set feature names
void	setFeatureTypes(String[] values) Set feature types
void	setGroup(int[] group) Set group sizes of DMatrix (used for ranking)
void	setLabel(Column column) Set label of DMatrix from array interface
void	setLabel(float[] labels) set label of dmatrix
void	setQueryId(Column column) Set query id of DMatrix from array interface
void	setQueryId(int[] qid) Set query ids (used for ranking)
void	setWeight(Column column) Set weight of DMatrix from array interface
void	setWeight(float[] weights) set weight of each instance
DMatrix	slice(int[] rowIndex) Slice the DMatrix and return a new DMatrix that only contains `rowIndex`.

Methods inherited from class java.lang.Object

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

Field Detail

handle

protected long handle

Constructor Detail

DMatrix

public DMatrix(Iterator<LabeledPoint> iter,
               String cacheInfo)
        throws XGBoostError

Create DMatrix from iterator.

Parameters:

iter - The data iterator of mini batch to provide the data.

cacheInfo - Cache path information, used for external memory setting, can be null.

Throws:

XGBoostError

DMatrix

public DMatrix(Iterator<LabeledPoint> iter,
               String cacheInfo,
               float missing)
        throws XGBoostError

Create DMatrix from iterator.

Parameters:

iter - The data iterator of mini batch to provide the data.

cacheInfo - Cache path information, used for external memory setting, can be null.

missing - the missing value

Throws:

XGBoostError

DMatrix

public DMatrix(String dataPath)
        throws XGBoostError

Create DMatrix by loading libsvm file from dataPath

Parameters:

dataPath - The path to the data.

Throws:

XGBoostError

DMatrix

@Deprecated
public DMatrix(long[] headers,
                           int[] indices,
                           float[] data,
                           DMatrix.SparseType st)
                    throws XGBoostError

Deprecated.

Create DMatrix from Sparse matrix in CSR/CSC format.

Parameters:

headers - The row index of the matrix.

indices - The indices of presenting entries.

data - The data content.

st - Type of sparsity.

Throws:

XGBoostError

DMatrix

public DMatrix(long[] headers,
               int[] indices,
               float[] data,
               DMatrix.SparseType st,
               int shapeParam)
        throws XGBoostError

Create DMatrix from Sparse matrix in CSR/CSC format.

Parameters:

headers - The row index of the matrix.

indices - The indices of presenting entries.

data - The data content.

st - Type of sparsity.

shapeParam - when st is CSR, it specifies the column number, otherwise it is taken as row number

Throws:

XGBoostError

DMatrix

public DMatrix(long[] headers,
               int[] indices,
               float[] data,
               DMatrix.SparseType st,
               int shapeParam,
               float missing,
               int nthread)
        throws XGBoostError

Throws:

XGBoostError

DMatrix

@Deprecated
public DMatrix(float[] data,
                           int nrow,
                           int ncol)
                    throws XGBoostError

Deprecated. Please specify the missing value explicitly using DMatrix(float[], int, int, float)

create DMatrix from dense matrix

Parameters:

data - data values

nrow - number of rows

ncol - number of columns

Throws:

XGBoostError - native error

DMatrix

public DMatrix(BigDenseMatrix matrix)
        throws XGBoostError

create DMatrix from a BigDenseMatrix

Parameters:

matrix - instance of BigDenseMatrix

Throws:

XGBoostError - native error

DMatrix

public DMatrix(float[] data,
               int nrow,
               int ncol,
               float missing)
        throws XGBoostError

create DMatrix from dense matrix

Parameters:

data - data values

nrow - number of rows

ncol - number of columns

missing - the specified value to represent the missing value

Throws:

XGBoostError

DMatrix

public DMatrix(BigDenseMatrix matrix,
               float missing)
        throws XGBoostError

create DMatrix from dense matrix

Parameters:

matrix - instance of BigDenseMatrix

missing - the specified value to represent the missing value

Throws:

XGBoostError

DMatrix

protected DMatrix(long handle)

used for DMatrix slice

DMatrix

public DMatrix(ColumnBatch columnBatch,
               float missing,
               int nthread)
        throws XGBoostError

Create the normal DMatrix from column array interface

Parameters:

columnBatch - the XGBoost ColumnBatch to provide the array interface of feature columns

missing - missing value

nthread - threads number

Throws:

XGBoostError

Method Detail

setQueryId

public void setQueryId(Column column)
                throws XGBoostError

Set query id of DMatrix from array interface

Parameters:

column - the XGBoost Column to provide the array interface of query id column

Throws:

XGBoostError - native error

getFeatureNames

public String[] getFeatureNames()
                         throws XGBoostError

Get feature names

Returns:

an array of feature names to be returned

Throws:

XGBoostError

setFeatureNames

public void setFeatureNames(String[] values)
                     throws XGBoostError

Set feature names

Parameters:

values - feature names to be set

Throws:

XGBoostError

getFeatureTypes

public String[] getFeatureTypes()
                         throws XGBoostError

Get feature types

Returns:

an array of feature types to be returned

Throws:

XGBoostError

setFeatureTypes

public void setFeatureTypes(String[] values)
                     throws XGBoostError

Set feature types

Parameters:

values - feature types to be set

Throws:

XGBoostError

getGroup

public int[] getGroup()
               throws XGBoostError

Get group sizes of DMatrix

Returns:

group size as array

Throws:

XGBoostError - native error

setGroup

public void setGroup(int[] group)
              throws XGBoostError

Set group sizes of DMatrix (used for ranking)

Parameters:

group - group size as array

Throws:

XGBoostError - native error

setQueryId

public void setQueryId(int[] qid)
                throws XGBoostError

Set query ids (used for ranking)

Parameters:

qid - the query ids

Throws:

XGBoostError - native error

getLabel

public float[] getLabel()
                 throws XGBoostError

get label values

Returns:

label

Throws:

XGBoostError - native error

setLabel

public void setLabel(Column column)
              throws XGBoostError

Set label of DMatrix from array interface

Parameters:

column - the XGBoost Column to provide the array interface of label column

Throws:

XGBoostError - native error

setLabel

public void setLabel(float[] labels)
              throws XGBoostError

set label of dmatrix

Parameters:

labels - labels

Throws:

XGBoostError - native error

getWeight

public float[] getWeight()
                  throws XGBoostError

get weight of the DMatrix

Returns:

weights

Throws:

XGBoostError - native error

setWeight

public void setWeight(Column column)
               throws XGBoostError

Set weight of DMatrix from array interface

Parameters:

column - the XGBoost Column to provide the array interface of weight column

Throws:

XGBoostError - native error

setWeight

public void setWeight(float[] weights)
               throws XGBoostError

set weight of each instance

Parameters:

weights - weights

Throws:

XGBoostError - native error

getBaseMargin

public float[] getBaseMargin()
                      throws XGBoostError

Get base margin of the DMatrix.

Throws:

XGBoostError

setBaseMargin

public void setBaseMargin(Column column)
                   throws XGBoostError

Set base margin of DMatrix from array interface

Parameters:

column - the XGBoost Column to provide the array interface of base margin column

Throws:

XGBoostError - native error

setBaseMargin

public void setBaseMargin(float[] baseMargin)
                   throws XGBoostError

Set base margin (initial prediction).

The margin must have the same number of elements as the number of rows in this matrix.

Throws:

XGBoostError

setBaseMargin

public void setBaseMargin(float[][] baseMargin)
                   throws XGBoostError

Set base margin (initial prediction).

Throws:

XGBoostError

slice

public DMatrix slice(int[] rowIndex)
              throws XGBoostError

Slice the DMatrix and return a new DMatrix that only contains `rowIndex`.

Parameters:

rowIndex - row index

Returns:

sliced new DMatrix

Throws:

XGBoostError - native error

rowNum

public long rowNum()
            throws XGBoostError

get the row number of DMatrix

Returns:

number of rows

Throws:

XGBoostError - native error

nonMissingNum

public long nonMissingNum()
                   throws XGBoostError

Get the number of non-missing values of DMatrix.

Returns:

The number of non-missing values

Throws:

XGBoostError - native error

saveBinary

public void saveBinary(String filePath)

save DMatrix to filePath

getHandle

public long getHandle()

Get the handle

finalize

protected void finalize()

Overrides:

finalize in class Object

dispose

public void dispose()

getQuantileCut

public DMatrix.QuantileCut getQuantileCut()
                                   throws XGBoostError

Get the Quantile Cut.

Returns:

QuantileCut

Throws:

XGBoostError