stabsel: Stability selection

In Rjacobucci/regsem: Regularized Structural Equation Modeling

Stability selection

Description

Stability selection

Usage

stabsel(
  data,
  model,
  det.range = FALSE,
  from,
  to,
  times = 50,
  jump = 0.01,
  detr.nlambda = 20,
  n.lambda = 40,
  n.boot = 100,
  det.thr = FALSE,
  p = 0.8,
  p.from = 0.5,
  p.to = 1,
  p.jump = 0.05,
  p.method = "aic",
  type = "lasso",
  pars_pen = "regressions",
  ...
)

Arguments

data	data frame
model	lavaan syntax model.
det.range	Whether to determine the range of penalization values for stability selection through bootstrapping. Default is FALSE, from and to arguments are then needed. If set to TRUE, then jump, times and detr.nlambda arguments will be needed.
from	Minimum value of penalization values for stability selection.
to	Maximum value of penalization values for stability selection.
times	Number of bootstrapping sample used to determine the range. Default is 50.
jump	Amount to increase penalization each iteration. Default is 0.01
detr.nlambda	Number of penalization values to test for determining range.
n.lambda	Number of penalization values to test for stability selection.
n.boot	Number of bootstrap samples needed for stability selection.
det.thr	Whether to determine the probability threshold value. Default is FALSE, p is then needed. If set to TRUE, p.from, p.to, p.method arguments will be needed.
p	Probability threshold: above which selection probability is the path kept in the modle. Default value is 0.8.
p.from	Lower bound of probability threshold to test. Default is 0.5.
p.to	Upper bound of probability threshold to test. Default is 1.
p.jump	Amount to increase threshold each iteration. Default is 0.05.
p.method	Which fit index to use to choose a final model?
type	Penalty type
pars_pen	Parameter indicators to penalize.
...	Any additional arguments to pass to regsem() or cv_regsem().

Examples

library(regsem)
# put variables on same scale for regsem
HS <- data.frame(scale(HolzingerSwineford1939[,7:15]))
mod <- '
f =~ 1*x1 + x2 + x3 + x4 + x5 + x6 + x7 + x8 + x9
x1 ~~ r1*x2;x1 ~~ r2*x3;x1 ~~ r3*x4;x1 ~~ r4*x5
'
outt = cfa(mod, HS)

stabsel.out = stabsel(data=HS,model=mod,det.range=TRUE,detr.nlambda=20,n.lambda=5,
                    n.boot=10,p=0.9,type="alasso", p.method="aic",
                    pars_pen=c("r1","r2","r3","r4"))
stabsel.out$selection_results

Rjacobucci/regsem documentation built on June 3, 2023, 5:20 a.m.