lm.mp: Massively parallel linear regression models
Description
Efficiently fits $V$ linear models with a common design matrix, where
$V$ may be very large, e.g., the number of voxels in a brain imaging
application.
Usage
lm.mp(Y, formula, store.fitted = FALSE)
Arguments
Y
$n \times V$ outcome matrix.
formula
a formula object such as "~ x1 + x2".
store.fitted
logical: Should the fitted values be stored? For large
$V$, setting this to TRUE may cause memory problems.
Value
coef
$p \times V$ matrix of coefficient estimates.
sigma2
$V$-dimensional vector of error variance estimates.
se.coef
$p \times V$ matrix of coefficient standard error
estimates.