stabplot is an R package designed to facilitate
regularisation tuning and convergence monitoring in stability selection
using Lasso. It provides two core functions, Regustab and
Convstab, which help visualise stability in regularised
models, supporting users in selecting appropriate regularisation
parameters and assessing convergence. Help functions are available
through R by ?stabplot, ?Regustab, and
?Convstab. Paper
The stabplot package depends on glmnet,
ggplot2, and latex2exp packages.
You can install the latest version of stabplot from
GitHub:
if (!require("devtools")){install.packages("devtools")} #installing devtools if it is not already installed
devtools::install_github("MahdiNouraie/stabplot") #installing stabplot
library(stabplot) #loading stabplotRegustab function creates a plot that displays stability
values in relation to regularisation values for Lasso through stability
selection. The plot highlights key lambda values, including
lambda.min, lambda.1se, and
lambda.stable. If lambda.stable is not
available, the function will display lambda.stable.1sd.
Regustab also returns a list containing the highlighted
regularisation values — min, 1se, and
stable (or stable.1sd) — accessible via
res$min, res$`1se`, and
res$stable.
A toy example of usage:
set.seed(123) # for reproducibility
x <- matrix(rnorm(1000), ncol = 10)
# create beta based on the first 3 columns of x and some error
beta <- c(1, 2, 3, rep(0, 7))
y <- x %*% beta + rnorm(100)
B <- 10 # number of sub-samples
res <- Regustab(x, y, B)
res
# $min
#[1] 0.07609021
#$`1se`
#[1] 0.2550241
#$stable
#[1] 0.3371269Convstab creates a plot displaying stability values
along with confidence intervals, against the sequential sub-sampling
index within stability selection. This plot aids in monitoring the
convergence status of stability values. The function uses
lambda.stable to generate the plot; if
lambda.stable is unavailable, it defaults to
lambda.stable.1sd.
The function returns a list containing the plot
(res$plot) and a data frame of variables whose selection
frequencies exceed thr (res$selected), so both can be
accessed or reused after the call.
A toy example of usage:
set.seed(123) # for reproducibility
x <- matrix(rnorm(1000), ncol = 10)
# create beta based on the first 3 columns of x and some error
beta <- c(0.5, 0.4, 0.3, rep(0, 7))
y <- x %*% beta + rnorm(100)
B <- 200 #number of sub-samples
alpha <- 0.05 #significance level of confidence interval
thr <- 0.5 # threshold for filtering variables based on their selection frequencies
res <- Convstab(x, y, B, alpha, thr)
res$selected
# Variable Selection_Frequency
#1 x1 0.970
#2 x2 0.895