Fit a grid of reference models

Using different values of "gold standard" measures of the outcome and number of expected non-zero parameters, fits a set of reference models via fit_reference.

Usage

grid_models(
  d,
  gss = c("Lvl.II (1)", "Lvl.II (2)"),
  N = 5,
  n_chosen = NA,
  bind = FALSE,
  ...
)

Arguments

d

Input data, e.g., generated by diagnose_pdd_sample.

gss

Gold standard variables.

N

Vector of expected non-zero coefficient for a regularised horseshoe prior.

n_chosen

Number of selected parameters.

bind

Should all combinations of gss and N be found (FALSE) or should they be simply binded (TRUE)

...

Other parameters for fit_reference().

Value

A list with four components:

grid

A grid of input values.

reference

List of reference models.

vs_object

List of variable selection results

vs_plots

Plots visualising variable selection

projection

List of selected models projected onto appropriate reference model

See also

• fit_reference() for reference model fitting

• perform_varsel() for variable selection implementation

• projpred::project() for projective prediction implementation