predict.composition

Object of class <code>composition</code> from the previous call of<code>fitcomp</code>.

object

newdata

If <code>"median"</code> (default), then the number of components is set to the median of the truncated negative binomial distribution, conditional on the estimated parameters; alternatively, it can be set to "mode."

n.method

Set the lower bound for predicted number of components. If <code>NULL</code> (default), then <code>l.bound</code> is taken from the <code>composition</code> object.

l.bound


Computes the predicted composition given a vector of covariates <code>newdata</code>.


 ~kwd1 

 ~kwd2 

Regression model where the response variable is a rank-indexed compositional vector (non-negative values that sum up to one and are ordered from the largest to the smallest). Parameters are estimated in the Bayesian framework using MCMC methods.

predict.composition: Predicted compositional vector

Description

Usage

Arguments

Value