Skip to content

Strip Linear Model Components

stripLMC.Rd

Strips linear model ("lm" or "glm") object components down the bare essentials to reduce size. Used typically when building thousands of models, e.g. during cross-validation to manage memory.

Usage

stripLMC(x)

Arguments

x

A model object, either lm or glm class.

Value

A stripped down (size) "lm" or "glm" object. For "lm" class, removes or sets -> 0 these elements: - model - fitted - assign - effects - xlevels For "glm" class, removes or sets to 0 these elements: - linear.predictors - terms - formula - family (family, link, linkfun, and linkinv remain)

See also

glm(), lm(), summary(), residuals()

Author

Stu Field

Examples

fat_lm <- function() {
  junk <- runif(1e5)
  stats::lm(vs ~ ., data = mtcars)
}
fat_glm <- function() {
  junk <- runif(1e5)   # excessive junk in envir
  stats::glm(vs ~ wt + disp, data = mtcars, family = "binomial")
}

# LM
lobstr::obj_size(fat_lm())
#> 823 kB
lobstr::obj_size(stripLMC(fat_lm()))
#> 17.94 kB

# GLM
lobstr::obj_size(fat_glm())
#> 883.91 kB
lobstr::obj_size(stripLMC(fat_glm()))
#> 37.74 kB