pls_filter {nestedcv}

Partial Least Squares filter

Description

Filter using coefficients from partial least squares (PLS) regression to select optimal predictors.

Usage

pls_filter(
  y,
  x,
  force_vars = NULL,
  nfilter,
  ncomp = 5,
  scale_x = TRUE,
  type = c("index", "names", "full"),
  ...
)

Arguments

y

Response vector

x

Matrix of predictors

force_vars

Vector of column names within x which are always retained in the model (i.e. not filtered). Default NULL means all predictors will be filtered.

nfilter

Either a single value for the total number of predictors to return. Or a vector of length ncomp to manually return predictors from each PLS component.

ncomp

the number of components to include in the PLS model.

scale_x

Logical whether to scale predictors before fitting the PLS model. This is recommended.

type

Type of vector returned. Default "index" returns indices, "names" returns predictor names, "full" returns a named vector of variable importance.

...

Other arguments passed to pls::plsr()

Details

The best predictors may overlap between components, so if nfilter is specified as a vector, the total number of unique predictors returned may be variable.

Value

Integer vector of indices of filtered parameters (type = "index") or character vector of names (type = "names") of filtered parameters. If type is "full" full output of coefficients from plsr is returned as a list for each model component ordered by highest absolute coefficient.