Nothing
R/likelihood.R
In bayescount: Power Calculations and Bayesian Analysis of Count Distributions and FECRT Data using MCMC
Defines functions
likelihood
Documented in
likelihood
likelihood <- function(model=stop("Please specify a distribution"), data=stop("Data must be specified"), mean=NA, variance=NA, zi=NA, shape=NA, scale=NA, iterations=min(1000, (length(as.matrix(mean)[,1])*length(shape))), log=TRUE, silent=FALSE, raw.output=FALSE){
model <- toupper(model)
## Cheat to stop function warning of non-zero inflated model when called from bayescount.single() and multiple deprecation warnings
zi.warn <- TRUE
if(inherits(silent, 'logical'))
if(silent=='TRUE')
silent <- TRUE
if(silent=='FALSE')
silent <- FALSE
if(silent=="noziwarnTRUE"){
zi.warn <- FALSE
silent <- TRUE
}
if(silent=="noziwarnFALSE"){
zi.warn <- FALSE
silent <- FALSE
}
## cheat to allow IP model to use SP model stuff:
if(model=="SP" | model=="ZISP" | model=="P" | model=="ZIP"){
if(!is.matrix(mean)){
oldmean <- mean
mean <- matrix(NA, ncol=length(na.omit(data)), nrow=length(oldmean))
mean[] <- oldmean
}
}
model <- switch(model, SP="P", IP="P", ZISP="ZIP", model)
models <- c("P", "ZIP", "G", "ZIG", "L", "ZIL", "W", "ZIW", "GP", "ZIGP", "LP", "ZILP", "WP", "ZIWP")
if((length(model) != 1) | sum(is.na(model)) > 0 | sum(model==models)!=1){
if(silent==FALSE){
cat("Invalid model selection. Please choose from ONE of the following distributions: ", sep="")
cat(models, sep=", ")
cat("\n")
}
stop("Invalid model selection")
}
suppressWarnings(if((sum(na.omit(strsplit(model, split=""))[[1]][] == c("Z", "I")))==2){
if(identical(zi, NA)){
stop("A value for zero-inflation (%) must be supplied when using a zero-inflated model")
}
zero.inflation <- TRUE
prob <- 1 - (zi/100)
model <- switch(model, ZIP="P", ZIG="G", ZIW="W", ZIL="L", ZIGP="GP", ZILP="LP", ZIWP="WP")
}else{
if(!is.na(zi)){
if(zi.warn==TRUE){
cat("A value for zero-inflation was provided for a non zero-inflated model. The appropriate zero-inflated model will be used instead of the model specified\n")
}
zero.inflation <- TRUE
prob <- 1 - (zi/100)
}else{
zero.inflation <- FALSE
}
}
)
chain.length <- length(as.matrix(mean)[,1]) * length(shape)
if(chain.length < iterations){
stop("The number of iterations to calculate the likelihood for must not be more than the length of the chain provided")
}
s.info <- Sys.info()
p.info <- .Platform
if(as.numeric(paste(R.version$major, R.version$minor, sep="")) < 26){
stop("Please update your version of R to the latest available (> 2.6.0 required)")
}
os <- p.info$OS.type
username <- as.character(s.info["user"])
rversion <- R.version$version
gui <- p.info$GUI
p.type <- p.info$pkgType
guitest <- c("R.GUI"=gui, "R.package.type"=p.type, "R.version"=list(rversion))
#if(guitest$os=='windows' | (guitest$R.GUI == "AQUA" & guitest$R.package.type == "mac.binary")) macgui <- TRUE else macgui <- FALSE
if(guitest$R.GUI == "AQUA" & guitest$R.package.type == "mac.binary") macgui <- TRUE else macgui <- FALSE
### The GUI version of R on mac currently doesn't support destructive backspace or carriage return, so the following is an alternative way of watching progress for Mac GUI
## later discovered that Windows doesn't handle them properly either - not sure about this now
sequence <- pmin(round((1:iterations) * (chain.length / iterations), digits=0), chain.length)
iters <- length(sequence)
progs <- character(length=iters)
progs[] <- ""
if(sum(is.na(data)) > 0){
if(silent==FALSE) cat("WARNING: Missing data was removed before calculating the likelihood\n")
data <- na.omit(data)
}
results <- matrix(NA, ncol=length(data), nrow=iters)
if(silent==FALSE){
cat("Calculating likelihood for the ", if(zero.inflation) "ZI", model, " model\n", sep="")
if(macgui==TRUE){
on.exit(cat("Finished calculating the likelihood\n"))
}else{
on.exit(cat("\b\rFinished calculating the likelihood \n"))
}
}
limits.data <- data
limits.data[limits.data==0] <- 1
integrate.min <- qgamma(0.001, shape=limits.data, rate=1)
integrate.max <- qgamma(0.999, shape=limits.data, rate=1)
integrate.min[data==0] <- 0
if(macgui==TRUE){
progress.function <- function(iteration){
cat(progs[iteration])
}
if(silent==FALSE && (model=="LP" | model=="GP" | model=="WP")){
cat("0%---------------------50%--------------------100%\n")
if(iterations >=50){
progs[pmin(round((1:50)*(iters/50), digits=0), iters)] <- "*"
}
if(iterations < 50 && iterations >= 25){
progs[pmin(round((1:25)*(iters/25), digits=0), iters)] <- "**"
}
if(iterations < 25 && iterations >= 10){
progs[pmin(round((1:10)*(iters/10), digits=0), iters)] <- "*****"
}
if(iterations < 10 && iterations >= 5){
progs[pmin(round((1:5)*(iters/5), digits=0), iters)] <- "**********"
}
if(iterations == 4){
progs[c(1,3)] <- "************"
progs[c(2,4)] <- "*************"
}
if(iterations == 3){
progs[c(1,3)] <- "*****************"
progs[2] <- "****************"
}
if(iterations == 2){
progs[] <- "*************************"
}
if(iterations == 1){
progs <- "**************************************************"
}
progs[iters] <- paste(progs[iters], "\n", sep="")
}
}else{
start.time <- Sys.time()
progress.function <- function(iteration){
post.time <- Sys.time()
total.time <- timestring(start.time, post.time, show.units=TRUE)
percent.complete <- iteration / iters
time.remaining <- timestring((as.integer(difftime(post.time, start.time, units="secs")) / percent.complete) - as.integer(difftime(post.time, start.time, units="secs")), show.units=TRUE)
if(iteration!=1){
cat("\b\r")
}
cat(round(percent.complete*100), "% complete, approximately ", time.remaining, " remaining ", sep="")
}
}
if(silent==TRUE){
progress.function <- function(iteration){
}
}
if(model=="P"){
if(sum(is.na(mean)) > 0){
stop("A value for mean must be supplied when using the Poisson model")
}
if(sum(!is.na(variance)) > 0 && silent==FALSE){
cat("The value for variance provided will be ignored using the Poisson model\n")
}
if(sum(!is.na(shape)) > 0 && silent==FALSE){
cat("The value for the shape parameter provided will be ignored using the Poisson model\n")
}
if(sum(!is.na(scale)) > 0 && silent==FALSE){
cat("The value for the scale parameter provided will be ignored using the Poisson model\n")
}
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log((((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * exp(-mean[sequence[iteration], datapoint])*mean[sequence[iteration], datapoint]^data[datapoint] / factorial(data[datapoint]))))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dpois(data[datapoint], mean[sequence[iteration], datapoint]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="L"){
if(sum(is.na(mean))>0){
stop("A value for mean must be supplied when using the lognormal model")
}
if(sum(is.na(variance))>0){
stop("A value for variance must be supplied when using the lognormal model")
}
if(sum(!is.na(shape))>0 && silent==FALSE){
cat("The value for the shape parameter provided will be ignored using the lognormal model\n")
}
if(sum(!is.na(scale))>0 && silent==FALSE){
cat("The value for the scale parameter provided will be ignored using the lognormal model\n")
}
newlnorms <- lnormal.params(mean, sqrt(variance))
lmean <- newlnorms[[1]]
lsd <- newlnorms[[2]]
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * dlnorm(data[datapoint], lmean[sequence[iteration]], lsd[sequence[iteration]])))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dlnorm(data[datapoint], lmean[sequence[iteration]], lsd[sequence[iteration]]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="LP"){
if(any(is.na(mean))){
stop("A value for mean must be supplied when using the lognormal Poisson model")
}
if(any(is.na(variance))){
stop("A value for variance must be supplied when using the lognormal Poisson model")
}
if(any(!is.na(shape)) && silent==FALSE){
cat("The value for the shape parameter provided will be ignored using the lognormal Poisson model\n")
}
if(any(!is.na(scale)) && silent==FALSE){
cat("The value for the scale parameter provided will be ignored using the lognormal Poisson model\n")
}
newlnorms <- lnormal.params(mean, sqrt(variance))
lmean <- newlnorms[[1]]
lsd <- newlnorms[[2]]
suppressWarnings({
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dlnorm(lambda, lmean[sequence[iteration]], lsd[sequence[iteration]])
results[iteration,datapoint] <- try(log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value)), silent=TRUE)
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dlnorm(lambda, lmean[sequence[iteration]], lsd[sequence[iteration]])
results[iteration,datapoint] <- try(log(integrate(f,integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value), silent=TRUE)
}
}
}
results <- matrix(data=as.numeric(results), nrow=length(results[,1]), ncol=length(results[1,]))})
likelihoods <- apply(results, 1, sum)
}
if(model=="G"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the gamma model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the gamma model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the gamma model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the gamma model\n")
}
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * dgamma(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dgamma(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="GP"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the gamma Poisson model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the gamma Poisson model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the gamma Poisson model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the gamma Poisson model\n")
}
suppressWarnings({
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dgamma(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value)), silent=TRUE)
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dgamma(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value), silent=TRUE)
}
}
}
results <- matrix(data=as.numeric(results), nrow=length(results[,1]), ncol=length(results[1,]))})
likelihoods <- apply(results, 1, sum)
}
if(model=="W"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the Weibull model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the Weibull model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the Weibull model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the Weibull model\n")
}
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * dweibull(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dweibull(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="WP"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the Weibull Poisson model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the Weibull Poisson model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the Weibull Poisson model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the Weibull Poisson model\n")
}
suppressWarnings({
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dweibull(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value)), silent=TRUE)
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dweibull(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value), silent=TRUE)
}
}
}
results <- matrix(data=as.numeric(results), nrow=length(results[,1]), ncol=length(results[1,]))})
# assign('gresults', results, pos=.GlobalEnv)
likelihoods <- apply(results, 1, sum)
}
if(log==FALSE){
likelihoods <- exp(likelihoods)
}
#if(silent==FALSE) cat("\n")
if(any(is.na(results)) && silent==FALSE && any(c(model=="WP", model=="GP", model=="LP"))) cat("Error: The likelihood could not be calculated at one or more iterations because an infinitely small likelihood was integrated for one or more datapoints at these iterations\n")
if(raw.output==TRUE){
return(list(likelihood=likelihoods, iteration=sequence))
}else{
if(length(likelihoods)==1){
return(likelihoods)
}else{
if(any(is.na(results))){
likeli.an <- c(l.95=NA, median=NA, u.95=NA, MAX=NA)
}else{
hpd <- coda::HPDinterval(coda::as.mcmc(likelihoods))
likeli.an <- c(l.95=hpd[1], median=median(likelihoods), u.95=hpd[2], MAX=max(likelihoods))
}
return(likeli.an)
}
}
}
Try the bayescount package in your browser
Any scripts or data that you put into this service are public.
bayescount documentation built on June 22, 2024, 12:24 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions likelihood
Documented in likelihood
likelihood <- function(model=stop("Please specify a distribution"), data=stop("Data must be specified"), mean=NA, variance=NA, zi=NA, shape=NA, scale=NA, iterations=min(1000, (length(as.matrix(mean)[,1])*length(shape))), log=TRUE, silent=FALSE, raw.output=FALSE){
model <- toupper(model)
## Cheat to stop function warning of non-zero inflated model when called from bayescount.single() and multiple deprecation warnings
zi.warn <- TRUE
if(inherits(silent, 'logical'))
if(silent=='TRUE')
silent <- TRUE
if(silent=='FALSE')
silent <- FALSE
if(silent=="noziwarnTRUE"){
zi.warn <- FALSE
silent <- TRUE
}
if(silent=="noziwarnFALSE"){
zi.warn <- FALSE
silent <- FALSE
}
## cheat to allow IP model to use SP model stuff:
if(model=="SP" | model=="ZISP" | model=="P" | model=="ZIP"){
if(!is.matrix(mean)){
oldmean <- mean
mean <- matrix(NA, ncol=length(na.omit(data)), nrow=length(oldmean))
mean[] <- oldmean
}
}
model <- switch(model, SP="P", IP="P", ZISP="ZIP", model)
models <- c("P", "ZIP", "G", "ZIG", "L", "ZIL", "W", "ZIW", "GP", "ZIGP", "LP", "ZILP", "WP", "ZIWP")
if((length(model) != 1) | sum(is.na(model)) > 0 | sum(model==models)!=1){
if(silent==FALSE){
cat("Invalid model selection. Please choose from ONE of the following distributions: ", sep="")
cat(models, sep=", ")
cat("\n")
}
stop("Invalid model selection")
}
suppressWarnings(if((sum(na.omit(strsplit(model, split=""))[[1]][] == c("Z", "I")))==2){
if(identical(zi, NA)){
stop("A value for zero-inflation (%) must be supplied when using a zero-inflated model")
}
zero.inflation <- TRUE
prob <- 1 - (zi/100)
model <- switch(model, ZIP="P", ZIG="G", ZIW="W", ZIL="L", ZIGP="GP", ZILP="LP", ZIWP="WP")
}else{
if(!is.na(zi)){
if(zi.warn==TRUE){
cat("A value for zero-inflation was provided for a non zero-inflated model. The appropriate zero-inflated model will be used instead of the model specified\n")
}
zero.inflation <- TRUE
prob <- 1 - (zi/100)
}else{
zero.inflation <- FALSE
}
}
)
chain.length <- length(as.matrix(mean)[,1]) * length(shape)
if(chain.length < iterations){
stop("The number of iterations to calculate the likelihood for must not be more than the length of the chain provided")
}
s.info <- Sys.info()
p.info <- .Platform
if(as.numeric(paste(R.version$major, R.version$minor, sep="")) < 26){
stop("Please update your version of R to the latest available (> 2.6.0 required)")
}
os <- p.info$OS.type
username <- as.character(s.info["user"])
rversion <- R.version$version
gui <- p.info$GUI
p.type <- p.info$pkgType
guitest <- c("R.GUI"=gui, "R.package.type"=p.type, "R.version"=list(rversion))
#if(guitest$os=='windows' | (guitest$R.GUI == "AQUA" & guitest$R.package.type == "mac.binary")) macgui <- TRUE else macgui <- FALSE
if(guitest$R.GUI == "AQUA" & guitest$R.package.type == "mac.binary") macgui <- TRUE else macgui <- FALSE
### The GUI version of R on mac currently doesn't support destructive backspace or carriage return, so the following is an alternative way of watching progress for Mac GUI
## later discovered that Windows doesn't handle them properly either - not sure about this now
sequence <- pmin(round((1:iterations) * (chain.length / iterations), digits=0), chain.length)
iters <- length(sequence)
progs <- character(length=iters)
progs[] <- ""
if(sum(is.na(data)) > 0){
if(silent==FALSE) cat("WARNING: Missing data was removed before calculating the likelihood\n")
data <- na.omit(data)
}
results <- matrix(NA, ncol=length(data), nrow=iters)
if(silent==FALSE){
cat("Calculating likelihood for the ", if(zero.inflation) "ZI", model, " model\n", sep="")
if(macgui==TRUE){
on.exit(cat("Finished calculating the likelihood\n"))
}else{
on.exit(cat("\b\rFinished calculating the likelihood \n"))
}
}
limits.data <- data
limits.data[limits.data==0] <- 1
integrate.min <- qgamma(0.001, shape=limits.data, rate=1)
integrate.max <- qgamma(0.999, shape=limits.data, rate=1)
integrate.min[data==0] <- 0
if(macgui==TRUE){
progress.function <- function(iteration){
cat(progs[iteration])
}
if(silent==FALSE && (model=="LP" | model=="GP" | model=="WP")){
cat("0%---------------------50%--------------------100%\n")
if(iterations >=50){
progs[pmin(round((1:50)*(iters/50), digits=0), iters)] <- "*"
}
if(iterations < 50 && iterations >= 25){
progs[pmin(round((1:25)*(iters/25), digits=0), iters)] <- "**"
}
if(iterations < 25 && iterations >= 10){
progs[pmin(round((1:10)*(iters/10), digits=0), iters)] <- "*****"
}
if(iterations < 10 && iterations >= 5){
progs[pmin(round((1:5)*(iters/5), digits=0), iters)] <- "**********"
}
if(iterations == 4){
progs[c(1,3)] <- "************"
progs[c(2,4)] <- "*************"
}
if(iterations == 3){
progs[c(1,3)] <- "*****************"
progs[2] <- "****************"
}
if(iterations == 2){
progs[] <- "*************************"
}
if(iterations == 1){
progs <- "**************************************************"
}
progs[iters] <- paste(progs[iters], "\n", sep="")
}
}else{
start.time <- Sys.time()
progress.function <- function(iteration){
post.time <- Sys.time()
total.time <- timestring(start.time, post.time, show.units=TRUE)
percent.complete <- iteration / iters
time.remaining <- timestring((as.integer(difftime(post.time, start.time, units="secs")) / percent.complete) - as.integer(difftime(post.time, start.time, units="secs")), show.units=TRUE)
if(iteration!=1){
cat("\b\r")
}
cat(round(percent.complete*100), "% complete, approximately ", time.remaining, " remaining ", sep="")
}
}
if(silent==TRUE){
progress.function <- function(iteration){
}
}
if(model=="P"){
if(sum(is.na(mean)) > 0){
stop("A value for mean must be supplied when using the Poisson model")
}
if(sum(!is.na(variance)) > 0 && silent==FALSE){
cat("The value for variance provided will be ignored using the Poisson model\n")
}
if(sum(!is.na(shape)) > 0 && silent==FALSE){
cat("The value for the shape parameter provided will be ignored using the Poisson model\n")
}
if(sum(!is.na(scale)) > 0 && silent==FALSE){
cat("The value for the scale parameter provided will be ignored using the Poisson model\n")
}
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log((((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * exp(-mean[sequence[iteration], datapoint])*mean[sequence[iteration], datapoint]^data[datapoint] / factorial(data[datapoint]))))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dpois(data[datapoint], mean[sequence[iteration], datapoint]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="L"){
if(sum(is.na(mean))>0){
stop("A value for mean must be supplied when using the lognormal model")
}
if(sum(is.na(variance))>0){
stop("A value for variance must be supplied when using the lognormal model")
}
if(sum(!is.na(shape))>0 && silent==FALSE){
cat("The value for the shape parameter provided will be ignored using the lognormal model\n")
}
if(sum(!is.na(scale))>0 && silent==FALSE){
cat("The value for the scale parameter provided will be ignored using the lognormal model\n")
}
newlnorms <- lnormal.params(mean, sqrt(variance))
lmean <- newlnorms[[1]]
lsd <- newlnorms[[2]]
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * dlnorm(data[datapoint], lmean[sequence[iteration]], lsd[sequence[iteration]])))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dlnorm(data[datapoint], lmean[sequence[iteration]], lsd[sequence[iteration]]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="LP"){
if(any(is.na(mean))){
stop("A value for mean must be supplied when using the lognormal Poisson model")
}
if(any(is.na(variance))){
stop("A value for variance must be supplied when using the lognormal Poisson model")
}
if(any(!is.na(shape)) && silent==FALSE){
cat("The value for the shape parameter provided will be ignored using the lognormal Poisson model\n")
}
if(any(!is.na(scale)) && silent==FALSE){
cat("The value for the scale parameter provided will be ignored using the lognormal Poisson model\n")
}
newlnorms <- lnormal.params(mean, sqrt(variance))
lmean <- newlnorms[[1]]
lsd <- newlnorms[[2]]
suppressWarnings({
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dlnorm(lambda, lmean[sequence[iteration]], lsd[sequence[iteration]])
results[iteration,datapoint] <- try(log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value)), silent=TRUE)
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dlnorm(lambda, lmean[sequence[iteration]], lsd[sequence[iteration]])
results[iteration,datapoint] <- try(log(integrate(f,integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value), silent=TRUE)
}
}
}
results <- matrix(data=as.numeric(results), nrow=length(results[,1]), ncol=length(results[1,]))})
likelihoods <- apply(results, 1, sum)
}
if(model=="G"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the gamma model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the gamma model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the gamma model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the gamma model\n")
}
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * dgamma(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dgamma(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="GP"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the gamma Poisson model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the gamma Poisson model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the gamma Poisson model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the gamma Poisson model\n")
}
suppressWarnings({
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dgamma(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value)), silent=TRUE)
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dgamma(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value), silent=TRUE)
}
}
}
results <- matrix(data=as.numeric(results), nrow=length(results[,1]), ncol=length(results[1,]))})
likelihoods <- apply(results, 1, sum)
}
if(model=="W"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the Weibull model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the Weibull model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the Weibull model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the Weibull model\n")
}
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * dweibull(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])))
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
results[iteration,datapoint] <- log(dweibull(data[datapoint], shape=shape[sequence[iteration]], scale=scale[sequence[iteration]]))
}
}
}
likelihoods <- apply(results, 1, sum)
}
if(model=="WP"){
if(any(is.na(shape))){
stop("A value for the shape parameter must be supplied when using the Weibull Poisson model")
}
if(any(is.na(scale))){
stop("A value for the scale parameter must be supplied when using the Weibull Poisson model")
}
if(any(!is.na(mean)) && silent==FALSE){
cat("The value for mean provided will be ignored using the Weibull Poisson model\n")
}
if(any(!is.na(variance)) && silent==FALSE){
cat("The value for variance provided will be ignored using the Weibull Poisson model\n")
}
suppressWarnings({
if(zero.inflation==TRUE){
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dweibull(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(((data[datapoint]==0) * (1 - prob[sequence[iteration]])) + (prob[sequence[iteration]] * integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value)), silent=TRUE)
}
}
}else{
for(iteration in 1:iters){
progress.function(iteration)
for(datapoint in 1:length(data)){
f <- function(lambda) dpois(data[datapoint], lambda) * dweibull(lambda, shape=shape[sequence[iteration]], scale=scale[sequence[iteration]])
results[iteration,datapoint] <- try(log(integrate(f, integrate.min[datapoint], integrate.max[datapoint], stop.on.error=FALSE)$value), silent=TRUE)
}
}
}
results <- matrix(data=as.numeric(results), nrow=length(results[,1]), ncol=length(results[1,]))})
# assign('gresults', results, pos=.GlobalEnv)
likelihoods <- apply(results, 1, sum)
}
if(log==FALSE){
likelihoods <- exp(likelihoods)
}
#if(silent==FALSE) cat("\n")
if(any(is.na(results)) && silent==FALSE && any(c(model=="WP", model=="GP", model=="LP"))) cat("Error: The likelihood could not be calculated at one or more iterations because an infinitely small likelihood was integrated for one or more datapoints at these iterations\n")
if(raw.output==TRUE){
return(list(likelihood=likelihoods, iteration=sequence))
}else{
if(length(likelihoods)==1){
return(likelihoods)
}else{
if(any(is.na(results))){
likeli.an <- c(l.95=NA, median=NA, u.95=NA, MAX=NA)
}else{
hpd <- coda::HPDinterval(coda::as.mcmc(likelihoods))
likeli.an <- c(l.95=hpd[1], median=median(likelihoods), u.95=hpd[2], MAX=max(likelihoods))
}
return(likeli.an)
}
}
}
Try the bayescount package in your browser
Any scripts or data that you put into this service are public.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.