lm_each_eclairs: Fit linear model on each feature after decorrelating

In decorrelate: Decorrelation Projection Scalable to High Dimensional Data

Fit linear model on each feature after decorrelating

Description

Fit linear model on each feature after applying decorrelation projection to response and predictors.

Usage

lm_each_eclairs(
  formula,
  data,
  X,
  ecl,
  subset,
  weights,
  na.action,
  method = "qr",
  model = TRUE,
  x = FALSE,
  y = FALSE,
  qr = TRUE,
  singular.ok = TRUE,
  contrasts = NULL,
  offset,
  ...
)

Arguments

formula	an object of class 'formula' (or one that can be coerced to that class): a symbolic description of the model to be fitted.
data	a matrix or data.frame containing the variables in the model
X	matrix or data.frame where each column stores a predictor to be evaluated by the regression model one at a time. The i^{th} model includes X[,i] as a predictor.
ecl	estimate of covariance/correlation matrix from eclairs storing U, d_1^2, \lambda and \nu
subset	same as for lm
weights	same as for lm
na.action	same as for lm
method	same as for lm
model	same as for lm
x	same as for lm
y	same as for lm
qr	same as for lm
singular.ok	same as for lm
contrasts	same as for lm
offset	same as for lm
...	other arguments passed to lm()

Value

data.frame with columns beta, se, tsat, pvalue storing results for regression model fit for each feature

Examples

library(Rfast)

n <- 800 # number of samples
p <- 200 # number of features

# create correlation matrix
Sigma <- autocorr.mat(p, .9)

# draw data from correlation matrix Sigma
Y <- rmvnorm(n, rep(0, p), sigma = Sigma * 5.1, seed = 1)

# eclairs decomposition
ecl <- eclairs(Y)

# simulate covariates
data <- data.frame(matrnorm(p, 2, seed = 1))
colnames(data) <- paste0("v", 1:2)

# simulate response
y <- rnorm(p)

# Simulate 1000 features to test
X <- matrnorm(p, 1000, seed = 1)
colnames(X) <- paste0("set_", seq(ncol(X)))

# Use linear model to test each feature stored as columns in X
res <- lm_each_eclairs(y ~ v1 + v2, data, X, ecl)

head(res)

# Analysis after non-linear transform
#------------------------------------

# Apply function to transforme data
f <- function(x) log(x^2 + 0.001)

# evaluate covariance of transformed data
ecl_transform <- cov_transform(ecl, f, 100)

# Use linear model to test each feature stored as columns in X
# in data transformed by f()
res2 <- lm_each_eclairs(f(y) ~ v1 + v2, data, X, ecl_transform)

head(res)

decorrelate documentation built on Aug. 8, 2025, 7:55 p.m.