estimate: Estimate parameters

In kevinmcgregor/mintalt: Learn Direct Interaction Networks (Edited version where lambda can be specified)

Description

This function performs iterative conditional modes to obtain maximum a posteriori estimates for β (covariate coefficients), w (latent abundances), and P (the precision matrix).

Usage

estimate(mfit)

Arguments

mfit	- a MInt model object.

Value

A MInt model object with the following attributes:

optim	List containing optimization details
optim$lambda	Value of the L1 penalty used during optimization
data	List containing the raw data
data$design	File path of the design matrix
data$response	File path of the response matrix
data$fmla	Formula used to model each response in terms of the design variables
data$y	Raw numerical data for the response matrix
data$xd	Design matrix in categorical form
data$x	Design matrix in numerical form
param	List containing parameter estimates
param$beta	p-covariates x o-responses matrix of regression coefficients
param$w	n-samples x o-responses matrix of latent abundances
param$P	o-responses x o-responses precision matrix

Examples

x <- system.file("extdata", "x.txt", package="MInt");
y <- system.file("extdata", "y.txt", package="MInt");
m <- mint(y,x,fmla = ~feature1 + feature2)
m <- estimate(m)

kevinmcgregor/mintalt documentation built on May 13, 2019, 4:34 p.m.