lm_dummy {TRexSelector}

Perform one random experiment

Description

Run one random experiment of the T-Rex selector (doi:10.48550/arXiv.2110.06048), i.e., generates dummies, appends them to the predictor matrix, and runs the forward selection algorithm until it is terminated after T_stop dummies have been selected.

Usage

lm_dummy(
  X,
  y,
  model_tlars,
  T_stop = 1,
  num_dummies = ncol(X),
  method = "trex",
  GVS_type = "IEN",
  type = "lar",
  corr_max = 0.5,
  lambda_2_lars = NULL,
  early_stop = TRUE,
  verbose = TRUE,
  intercept = FALSE,
  standardize = TRUE
)

Arguments

X

Real valued predictor matrix.

y

Response vector.

model_tlars

Object of the class tlars_cpp. It contains all state variables of the previous T-LARS step (necessary for warm-starts, i.e., restarting the forward selection process exactly where it was previously terminated).

T_stop

Number of included dummies after which the random experiments (i.e., forward selection processes) are stopped.

num_dummies

Number of dummies that are appended to the predictor matrix.

method

'trex' for the T-Rex selector (doi:10.48550/arXiv.2110.06048), 'trex+GVS' for the T-Rex+GVS selector (doi:10.23919/EUSIPCO55093.2022.9909883), 'trex+DA+AR1' for the T-Rex+DA+AR1 selector, 'trex+DA+equi' for the T-Rex+DA+equi selector, 'trex+DA+BT' for the T-Rex+DA+BT selector (doi:10.48550/arXiv.2401.15796), 'trex+DA+NN' for the T-Rex+DA+NN selector (doi:10.48550/arXiv.2401.15139).

GVS_type

'IEN' for the Informed Elastic Net (doi:10.1109/CAMSAP58249.2023.10403489), 'EN' for the ordinary Elastic Net (doi:10.1111/j.1467-9868.2005.00503.x).

type

'lar' for 'LARS' and 'lasso' for Lasso.

corr_max

Maximum allowed correlation between any two predictors from different clusters.

lambda_2_lars

lambda_2-value for LARS-based Elastic Net.

early_stop

Logical. If TRUE, then the forward selection process is stopped after T_stop dummies have been included. Otherwise the entire solution path is computed.

verbose

Logical. If TRUE progress in computations is shown when performing T-LARS steps on the created model.

intercept

Logical. If TRUE an intercept is included.

standardize

Logical. If TRUE the predictors are standardized and the response is centered.

Value

Object of the class tlars_cpp.

Examples

set.seed(123)
eps <- .Machine$double.eps
n <- 75
p <- 100
X <- matrix(stats::rnorm(n * p), nrow = n, ncol = p)
beta <- c(rep(3, times = 3), rep(0, times = 97))
y <- X %*% beta + rnorm(n)
res <- lm_dummy(X = X, y = y, T_stop = 1, num_dummies = 5 * p)
beta_hat <- res$get_beta()[seq(p)]
support <- abs(beta_hat) > eps
support