qrBlockApply: Blockwise Cholesky Factorization

In biglars: Scalable Least-Angle Regression and Lasso

Description

Cholesky factorization of a crossproduct matrix via blockwise orthogonal transformation.

Usage

qrBlockApply(x, y = NULL, blockSize = NULL)

Arguments

x	A numeric matrix or ff. The number of columns should not be larger than the number of rows.
y	An optional numeric vector or 1 column ff of responses.
blockSize	This is used for setting the block size for sequential access. The default is determined by methods from class ff.

Value

A list with the following elements:

R	Upper triangular Cholesky factor of crossprod(x).
Qty	The corresponding transformation of y.

References

C. Fraley and T. Hesterberg (2007), "Least-Angle Regression for Large Datasets", Technical Report, Insightful Corporation.

See Also

biglars.fit

Examples

data(diabetes)

qrx <- qr(diabetes$x)

list(R = qr.R(qrx), Qty = crossprod(qr.Q(qrx), diabetes$y))

qrBlockApply( diabetes$x, diabetes$y, blockSize = 34)

biglars documentation built on May 2, 2019, 3:08 a.m.