bnlogl {bernor}R Documentation

Observed Data Log Likelihood for Bernoulli-Normal Random Effects Model

Description

Evaluate by Monte Carlo the observed data log likelihood for Bernoulli regression model with normal random effects.

Usage

bnlogl(y, beta, sigma, nmiss, x, z, i, model, deriv = 0, weigh)

Arguments

y a zero-one-valued (Bernoulli) matrix, the response.
beta the fixed effect vector.
sigma the scale parameter vector for the random effects.
nmiss integer, the number of simulations of the missing data.
x the model matrix for fixed effects.
z the model matrix for random effects.
i the index vector for random effects.
model the model for the importance sampling distribution, an object of class model produced by the model function.
deriv the number of derivatives wanted. No more than 3. Zero, the default, means no derivatives. Three is a kludge. It doesn't mean third derivatives but to output ``big V hat''.
weigh weights. Positive integer valued vector of length ncol(y). May be missing in which case all weights one is assumed.

Details

evaluates by good old-fashioned (IID) Monte Carlo observed data log density as if doing the R statements 

    logf <- rep(NA, nmiss)
    nobs <- ncol(y) 
    save.Random.seed <- .Random.seed
    for (j in 1:nmiss) {
        .Random.seed <<- save.Random.seed
        logf[j] <- bnmarg(y[ , j], beta, b, sigma, x, z, i)$value
    }
    sum(weigh * logf)

Value

A list containing some of the following components:

value the function value.
gradient the gradient vector. The length is nparm, which is length(beta) + length(mu).
hessian the hessian matrix. The dimension is nparm by nparm.
bigv the ``big V hat'' matrix. The dimension is nparm by nparm.

See Also

bernor, bnmarg.

Examples

data(salam)
attach(salam)
beta <- c(0.91, -3.01, -0.49, 3.54)
sigma <- c(1.18, 0.98)
moo <- model("gauss", length(i), 1)
nmiss <- 100
bnlogl(y, beta, sigma, nmiss, x, z, i, moo, deriv = 3)
[Package bernor version 0.3-6 Index]