VLMCX (version 1.0)

BIC: Bayesian Information Criteria for for VLMCX objects that compose Variable Length Markov Chains with Exogenous Covariates

Description

Computes the Bayesian Information Criteria for the data using the estimated parameters of the multinomial logistic regression in the VLMCX fit.

Usage

BIC(fit)

Value

a numeric value with the corresponding BIC.

Arguments

fit

a betaVLMC object.

Author

Adriano Zanin Zambom <[e-mail]>

Examples

# \donttest{
set.seed(1)
n = 1000
d = 2

X = cbind(rnorm(n), rnorm(n))
p = 1/(1 + exp(0.5 + -2*X[,1] - 3.5*X[,2]))

y = c(sample(1:0,1), rbinom(n,1, p)) 

fit = maximum.context(y[1:n], X, max.depth = 3, n.min = 25)
draw(fit)
BIC(fit)
##[1] 696.0343

fit = VLMCX(y[1:n], X, alpha.level = 0.001, max.depth = 3, n.min = 25)
draw(fit)
BIC(fit)
##[1] 588.9432
# }