Runs the Gauss Dantzig Selector (GDS) multiple times, each time
with a different set of randomly selected two-factor interactions.
All m main effects are included in each GDS run. For each set of
randomly selected interactions, the best GDS output is chosen among
delta.n values of delta. We use kmeans with 2
clusters and BIC to select such best model.
StepI_chooseints(
delta.n = 10,
nint,
nrep,
Xmain,
Xint,
Y,
opt.heredity = c("none")
)A list containing the (a) matrix of the output of each GDS run with each row representing the selected effects from the corresponding GDS run, (b) a vector with the corresponding BIC values of each model.
a positive integer suggesting the number of delta values
to be tried. delta.n equally spaced values of delta will be used
strictly between 0 and max(|t(X)y|). The default value is set to 10.
a positive integer representing the number of randomly
chosen interactions. The suggested value to use is the ceiling of 20%
of the total number of interactions, that is, for m factors, we have
ceiling(0.2(m choose 2)).
a positive integer representing the number of times GDS should
be run. The suggested value is (m choose 2).
a \(n \times m\) matrix of m main effects.
a matrix of \({m \choose 2}\) two-factor interactions.
a vector of n responses.
a string with either none, or weak, or strong. Denotes
whether the effect-heredity (weak or strong) should be embedded in GDS-ARM.
The default value is none as suggested in Singh and Stufken (2022).