grpintprob.r

这个包里面含有利用极大似然估计方法来拟合成组数据和条件数据的有限混合分布模型的函数

## last modified June 2002

grpintprob <- function(mixdat, mixpar, dist, constr) 
{
    m <- nrow(mixdat)
    k <- nrow(mixpar)
    mu <- mixpar[, 2]
    sigma <- mixpar[, 3]
    if (dist == "norm") {
        par1 <- mu
        par2 <- sigma
        mixcdf <- t(sapply(mixdat[-m, 1], pnorm, par1, par2))
    }
    else if (dist == "lnorm") {
        par2 <- sqrt(log((sigma/mu)^2 + 1))
        par1 <- log(mu) - (par2^2)/2
        mixcdf <- t(sapply(mixdat[-m, 1], plnorm, par1, par2))
    }
    else if (dist == "gamma") {
        par1 <- (mu/sigma)^2
        par2 <- mu/(sigma^2)
        mixcdf <- t(sapply(mixdat[-m, 1], pgamma, par1, par2))
    }
    else if (dist == "weibull") {
        par <- weibullpar(mu, sigma)
        par1 <- par$shape
        par2 <- par$scale
        mixcdf <- t(sapply(mixdat[-m, 1], pweibull, par1, par2))
    }
    else if (dist == "binom") {
        par1 <- constr$size
        par2 <- mu/constr$size
        mixcdf <- t(sapply(mixdat[-m, 1], pbinom, par1, par2))
    }
    else if (dist == "nbinom") {
        if (constr$consigma == "NBINOM") 
            par1 <- constr$size
        else par1 <- mu^2/(sigma^2 - mu)
        mixcdf <- t(sapply(mixdat[-m, 1], pnbinom, par1, mu = mu))
    }
    else if (dist == "pois") {
        par <- mu
        mixcdf <- t(sapply(mixdat[-m, 1], ppois, par))
    }
    if (k == 1) 
        mixcdf <- t(mixcdf)
    rbind(mixcdf, 1) - rbind(0, mixcdf)
}