sparse_linear_summary: Sparse linear summaries

In spencerwoody/possum: Model interpretation using using lower-dimensional posterior summarization

Sparse linear summaries

Description

Sparse linear summary

Usage

sparse_linear_summary(
  X,
  fhatmat = X %*% betaSamples,
  betaSamples,
  sigma2Samples = NA,
  adaptive = TRUE,
  varnames = NA,
  alpha = 0.05,
  ...
)

Arguments

X	N \times p design matrix
fhatmat	N \times NMC matrix of posterior draws of the function f, where N is the number of observations and NMC is the number of Monte Carlo posterior samples. The user must specify fhatmat OR betaSamples
betaSamples	p \times NMC matrix of posterior draws of the coefficients for the (generalized) linear model
sigma2Samples	Optional vector of posterior samples for the residual variance (for a linear model)
adaptive	if TRUE (default), use adaptive lasso, weighting by the posterior mean. See Hahn and Carvalho (2015)
varnames	Optional vector of variable names
alpha	Return the alpha/2 and 1-alpha/2 posterior credible intervals for the summary
...	other arguments, e.g., to glmnet

Details

Compute a sparse linear summary of a nonparametric regression model or high-dimensional (generalized) linear model

Author(s)

Spencer Woody

spencerwoody/possum documentation built on Aug. 5, 2022, 12:18 a.m.