methods

Methods for 'lmSubsets' and 'lmSelect' Objects

Extractor methods for "lmSubsets" and "lmSelect" objects.

Usage

# S3 method for lmSubsets
variable.names(object, size, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSelect
variable.names(object, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSubsets
formula(x, size, best = 1, …)
# S3 method for lmSelect
formula(x, best, …)
# S3 method for lmSubsets
model.frame(formula, …)
# S3 method for lmSelect
model.frame(formula, …)
# S3 method for lmSubsets
model.matrix(object, size, best = 1, …)
# S3 method for lmSelect
model.matrix(object, best, …)
# S3 method for lmSubsets
model_response(data, …)
# S3 method for lmSelect
model_response(data, …)
# S3 method for lmSubsets
refit(object, size, best = 1, …)
# S3 method for lmSelect
refit(object, best = 1, …)
# S3 method for lmSubsets
deviance(object, size, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSelect
deviance(object, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSubsets
sigma(object, size, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSelect
sigma(object, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSubsets
logLik(object, size, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSelect
logLik(object, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSubsets
AIC(object, size, best = 1, …, k = 2, na.rm = TRUE, drop = TRUE)
# S3 method for lmSelect
AIC(object, best = 1, …, k = 2, na.rm = TRUE, drop = TRUE)
# S3 method for lmSubsets
BIC(object, size, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSelect
BIC(object, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSubsets
coef(object, size, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSelect
coef(object, best = 1, …, na.rm = TRUE, drop = TRUE)
# S3 method for lmSubsets
vcov(object, size, best = 1, …)
# S3 method for lmSelect
vcov(object, best = 1, …)
# S3 method for lmSubsets
fitted(object, size, best = 1, …)
# S3 method for lmSelect
fitted(object, best = 1, …)
# S3 method for lmSubsets
residuals(object, size, best = 1, …)
# S3 method for lmSelect
residuals(object, best = 1, …)

Details

The extractor methods work for "lmSubsets" and "lmSelect" objects, i.e., objects that have been generated using the formula interface.

If drop == FALSE, the extractor methods variable.names, deviance, sigma, logLik, AIC, BIC and coef return a data.frame object. If drop == TRUE, the return value is a logical matrix (variable.names), a numeric matrix (coef), or a numeric vector. If the drop parameter is not set explicitly when calling variable.names or coef, one-dimensional values are represented in a compact form.

If a desired extractor function is not available, refit can be called explicitly to obtain the corresponding "lm" object.

See Also

lmSubsets, lmSelect, refit.

Aliases

• methods
• model_response
• variable.names.lmSubsets
• formula.lmSubsets
• model.frame.lmSubsets
• model.matrix.lmSubsets
• model_response.lmSubsets
• refit.lmSubsets
• deviance.lmSubsets
• sigma.lmSubsets
• logLik.lmSubsets
• AIC.lmSubsets
• BIC.lmSubsets
• coef.lmSubsets
• vcov.lmSubsets
• fitted.lmSubsets
• residuals.lmSubsets
• variable.names.lmSelect
• formula.lmSelect
• model.frame.lmSelect
• model.matrix.lmSelect
• model_response.lmSelect
• refit.lmSelect
• deviance.lmSelect
• sigma.lmSelect
• logLik.lmSelect
• AIC.lmSelect
• BIC.lmSelect
• coef.lmSelect
• vcov.lmSelect
• fitted.lmSelect
• residuals.lmSelect

Examples

# NOT RUN {
## load data
data("AirPollution", package = "lmSubsets")

## fit subsets (5 best subsets per size)
lm_all <- lmSubsets(mortality ~ ., data = AirPollution, nbest = 5)

## extract information (for best submodel of size 3)
coef(lm_all, size = 3)
vcov(lm_all, size = 3)
residuals(lm_all, size = 3)
fitted(lm_all, size = 3)
model.matrix(lm_all, size = 3)

## select best (BIC) submodels
lm_best <- lmSelect(lm_all)

## extract information
deviance(lm_best)
logLik(lm_best)
AIC(lm_best)
BIC(lm_best, best = 1:5)

## refit model
lm5 <- refit(lm_all, size = 5)
summary(lm5)
## (Note that the p-values are not valid due to model selection.)
# }