WpProj (version 0.2.1)

WPSW: p-Wasserstein Distance Linear Projections Using a Stepwise Method

Description

p-Wasserstein Distance Linear Projections Using a Stepwise Method

Usage

WPSW(
  X,
  Y,
  theta,
  power = 2,
  force = NULL,
  direction = c("backward", "forward"),
  method = c("selection.variable", "scale", "projection"),
  transport.method = transport_options(),
  OTmaxit = 100,
  epsilon = 0.05,
  calc.theta = TRUE,
  model.size = NULL,
  parallel = NULL,
  display.progress = FALSE,
  ...
)

Value

An object of class WpProj

Arguments

X

matrix of covariates

Y

matrix of predictions

theta

optional parameter matrix for selection methods.

power

Power of the Wasserstein distance

force

Any covariates to force into the model?

direction

forward or backward selection

method

"selection.variable" or "projection

transport.method

Method for calculating the Wasserstein distance. Should be one of the outputs of transport_options().

OTmaxit

maximum number of iterations for the opt?imal transport methods

epsilon

hyperparameter if using sinkhorn iterations to approximate OT

calc.theta

should we get the linear coefficients

model.size

Maximum model size

parallel

foreach backend

display.progress

Display intermediate progress