EigenPrism: EigenPrism

In wal615/prime.total.effect: Total effect estimation of the environmental mixtures

Description

EigenPrism calculate the total effect based on a paper called EigenPrism: Inference for High-Dimensional Signal-to-Noise Ratios. The source code of this function could be found at the author's website.

Usage

EigenPrism(
  y,
  x,
  invsqrtSig = NULL,
  alpha = 0.05,
  target = c("beta2", "r2")[1],
  zero.ind = c(),
  diagnostics = F
)

Arguments

x:	n by p design matrix; columns will automatically be centered and scaled to variance 1
y:	response vector of length n (will automatically be centered) should not contain intercept column, since both y and x will be centered
invsqrtSig:	if columns of x not independent, p by p positive definite matrix which is the square-root of the inverse of Sig, where Sig is the *correlation* matrix of the x (default is identity)
alpha:	significance level for confidence interval (default = 0.05)
target:	target of estimation/inference beta2 Total effect, which is the squared 2-norm of the coefficient vector: sum(beta^2). r2 Estimated the proportion of the variance explained x, which is the fraction of variance of y explained by x
zero.ind:	vector of which indices of the weight vector w to constrain to zero (default is none)
diagnostics:	boolean (default = T) for whether to generate diagnostic plots for the V.i, lambda.i, and w.i

Value

1. beta2 or r2 estimate: unbiased estimate of the target (for beta2 or r2)

2. CI: 100*(1-alpha)

Author(s)

Author: Lucas Janson (statweb.stanford.edu/~ljanson)

wal615/prime.total.effect documentation built on April 29, 2020, 2:05 p.m.