get_stability_paths: Extract stability paths
In fabian-s/stabpath: compute stability selection paths

Description Usage Arguments Details Value Examples

Extract stability paths from refits of a model object on resampled or subsampled data.

1 2	get_stability_paths(model, data, reps = 100, method = "subsample", strata = NULL, fraction = 0.5)

`model`	a fitted model object for which `refit` and `get_selected` methods are defined
`data`	the data used for the model fit, a `data.frame` or a list containing `data.frame`s, vectors and/or matrices. See Details.
`reps`	how many refits of `model` on resampled `data` to perform
`method`	resampling method, either "subsample" (without replacement) or "bootstrap" (with replacement)
`strata`	for subsampling, a vector with length nrow(data) defining the strata
`fraction`	subsampling fraction

List-shaped data arguments are necessary a) for models that use weights or offsets and/or b) for modeling functions that receive the design matrix and the response vector as arguments directly instead of a model formula and a corresponding dataset. In these cases data has to be a named list with names corresponding to the modelling function's respective arguments (for example, list(x = <design matrix>, y = <response>, weights = <model weights>, offset = <model offsets>) or list(data = <dataframe>, weights = <model weights>))

a <max. subsetsize + 1> x <covariates> matrix of relative selection frequencies with class stabpath. First row is the null model (no covariates, i.e., all 0s)

library(ElemStatLearn)
library(leaps)
data(prostate)
data <- prostate
max_formula <- lpsa ~ (. - train)
model <- regsubsets(max_formula, data=data, nbest = 1, nvmax = 8,
  really.big = TRUE)
set.seed(20141020)
get_stability_paths(model, data, reps=10)