sgpls: Sparse Generalized Projection to Latent Structures

In abnormally-distributed/cvreg: Cross Validation and Robust Estimation Utilities

Description

Sparse projection to latent structures (partial least squares) is an extension of PLS that takes advantage of L1 regularization via a LARS-like algorithm to select predictors. Predictors which do not load highly on any of the latent variables get dropped from the model, and the corresponding regression estimates are shrunk to zero. Here sparse PLS is extended to include generalized linear models. Only univariate outcomes are supported here.

Usage

sgpls(
  formula,
  data,
  ncomp = 2,
  lambda = 0.01,
  family = "gaussian",
  link = "identity",
  scale = T
)

Arguments

formula	model formula
data	a data frame
ncomp	number of components to retain
lambda	a regularization parameter.
family	"gaussian", "poisson", "negative.binomial", "binomial", "Gamma", or "inverse.gaussian"
link	the link function. see details for available options.

References

Chun, H., & Keles, S. (2010). Sparse partial least squares regression for simultaneous dimension reduction and variable selection. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 72(1), 3–25. doi:10.1111/j.1467-9868.2009.00723.x

abnormally-distributed/cvreg documentation built on May 3, 2020, 3:45 p.m.