R/gMLE.pg.R

In BayesGOF: Bayesian Modeling via Frequentist Goodness-of-Fit

gMLE.pg <-
function(cnt.vec, exposure = NULL, start.par = c(1,1)){
    if (is.null(exposure) == TRUE) {
        exposure <- rep(1, length(cnt.vec))
    }
    else {
        exposure <- exposure
    }
    logL <- function(par, x, e) {
        f <- choose(x + par[1] - 1, x) * (par[2]/(e + par[2]))^par[1] * 
            (e/(e + par[2]))^x
        -sum(log(f))
    }
    suppressWarnings(par.g <- optim(start.par, logL, x = cnt.vec, 
        e = exposure, hessian = TRUE)$par)
    par.g[2] <- 1/par.g[2]
    return(par.g)
}