getPredCov: Get the predicted covariance matrix based on the 'micore'...

In kevinmcgregor/micore: Microbiome Covariance Regression

Description

Get the predicted covariance matrix based on the micore model fit. Can also return correlation, precision, and partial correlation matrices.

Usage

getPredCov(
  obj,
  newdata = NULL,
  quant = c(0.025, 0.975),
  type = c("cov", "cor", "prec", "pcor"),
  post.stat = c("mean", "median")
)

Arguments

obj	An object of class micore.
newdata	n optional numeric matrix containing covariates for new observations to get predicted covariance matrices for.
quant	Numeric vector specifying the quantiles of the posterior to return for predicted covariances.
type	Type of matrix to return: covariance, correlation, precision, or partial correlation.
post.stat	Character specifying whether the predictions be based on the posterior mean or median.

Value

A list containing:

• fit: A matrix containing the posterior mean (or median) of the covariance matrix.

• quant: A list containing the requested quantiles from the posterior mean for the covariance matrix.

Examples

n <- 50
p <- 5
X <- cbind(1, rnorm(n))
counts <- matrix(0, n, p+1)
for (i in 1:n) {
  counts[i,] <- rmultinom(1, size=100, prob=rep(1,p+1))
}

library(micore)
mc.fit <- micore(counts, X, n.samp=100, n.burn=100, n.chain=1)

new.dat <- cbind(c(1,1,1),c(0,1,0))

c.mat <- getPredCov(mc.fit, new.dat)
pc.mat <- getPredCov(mc.fit, new.dat, type="pcor")

kevinmcgregor/micore documentation built on June 9, 2021, 10:29 p.m.