hex.tree.xgboost.matrix

Class SparseMatrix

java.lang.Object

hex.tree.xgboost.matrix.SparseMatrix

public final class SparseMatrix
extends java.lang.Object

Sparse Matrix representation for XGBoost CSR: long[] rowHeaders = new long[] {0, 2, 4, 7}; // offsets float[] data = new float[] {1f,2f, 4f,3f, 3f,1f,2f}; // non-zeros across each row int[] colIndex = new int[] {0, 2, 0, 3, 0, 1, 2}; // col index for each non-zero

Field Summary

Fields

Modifier and Type	Field and Description
int[][]	_colIndices
long[][]	_rowHeaders
float[][]	_sparseData
static int	MAX_DIM Maximum size of one dimension of SPARSE matrix with data.

Constructor Summary

Constructors

Constructor and Description
SparseMatrix(float[][] sparseData, long[][] rowIndices, int[][] colIndices) Constructs a SparseMatrix instance

Method Summary

Methods inherited from class java.lang.Object

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

Field Detail

MAX_DIM

public static int MAX_DIM

Maximum size of one dimension of SPARSE matrix with data. Sparse matrix is square matrix.

_sparseData

public final float[][] _sparseData

_rowHeaders

public final long[][] _rowHeaders

_colIndices

public final int[][] _colIndices

Constructor Detail

SparseMatrix

public SparseMatrix(float[][] sparseData,
                    long[][] rowIndices,
                    int[][] colIndices)

Constructs a SparseMatrix instance

Parameters:

sparseData - Non-zero data of a sparse matrix

rowIndices - Indices to elements in sparseData rows begin with

colIndices - Column indices of elements in sparseData