phenology (version 10.3)

phenology_MHmcmc: Run the Metropolis-Hastings algorithm for data

Description

Run the Metropolis-Hastings algorithm for data.
The number of iterations is n.iter+n.adapt+1 because the initial likelihood is also displayed.
I recommend thin=10.
If initial point is maximum likelihood, n.adapt = 0 is a good solution.
The parameters intermediate and filename are used to save intermediate results every 'intermediate' iterations (for example 1000). Results are saved in a file of name filename.
The parameter previous is used to indicate the list that has been save using the parameters intermediate and filename. It permits to continue a mcmc search.
These options are used to prevent the consequences of computer crash or if the run is very very long and computer processes are time limited.

Usage

phenology_MHmcmc(
  result = stop("An output from fit_phenology() must be provided"),
  n.iter = 10000,
  parametersMCMC = stop("A model generated with phenology_MHmcmc_p() must be provided"),
  n.chains = 1,
  n.adapt = 1000,
  thin = 1,
  WAIC = FALSE,
  WAIC.bybeach = TRUE,
  trace = FALSE,
  traceML = FALSE,
  adaptive = TRUE,
  adaptive.lag = 500,
  adaptive.fun = function(x) {
     ifelse(x > 0.234, 1.3, 0.7)
 },
  intermediate = NULL,
  filename = "intermediate.Rdata",
  previous = NULL
)

Value

A list with resultMCMC being mcmc.list object, resultLnL being likelihoods and parametersMCMC being the parameters used

Arguments

result

An object obtained after a SearchR fit

n.iter

Number of iterations for each step

parametersMCMC

A set of parameters used as initial point for searching with information on priors

n.chains

Number of replicates

n.adapt

Number of iterations before to store outputs

thin

Number of iterations between each stored output

WAIC

Should WAIC information be saved?

WAIC.bybeach

Should the WAIC be saved by beach or by count?

trace

TRUE or FALSE or period, shows progress

traceML

TRUE or FALSE to show ML

adaptive

Should an adaptive process for SDProp be used

adaptive.lag

Lag to analyze the SDProp value in an adaptive content

adaptive.fun

Function used to change the SDProp

intermediate

Period for saving intermediate result, NULL for no save

filename

If intermediate is not NULL, save intermediate result in this file

previous

Previous result to be continued. Can be the filename in which intermediate results are saved.

Author

Marc Girondot

Details

phenology_MHmcmc runs the Metropolis-Hastings algorithm for data (Bayesian MCMC)

See Also

Other Phenology model: AutoFitPhenology(), BE_to_LBLE(), Gratiot, LBLE_to_BE(), LBLE_to_L(), L_to_LBLE(), MarineTurtles_2002, MinBMinE_to_Min(), adapt_parameters(), add_SE(), add_phenology(), extract_result(), fit_phenology(), likelihood_phenology(), logLik.phenology(), map_Gratiot, map_phenology(), par_init(), phenology(), phenology2fitRMU(), phenology_MHmcmc_p(), plot.phenology(), plot.phenologymap(), plot_delta(), plot_phi(), print.phenology(), print.phenologymap(), print.phenologyout(), remove_site(), result_Gratiot, result_Gratiot1, result_Gratiot2, result_Gratiot_Flat, summary.phenology(), summary.phenologymap(), summary.phenologyout()

Examples

if (FALSE) {
library(phenology)
data(Gratiot)
# Generate a formatted list named data_Gratiot 
data_Gratiot <- add_phenology(Gratiot, name="Complete", 
    reference=as.Date("2001-01-01"), format="%d/%m/%Y")
# Generate initial points for the optimisation
parg <- par_init(data_Gratiot, fixed.parameters=NULL)
# Run the optimisation
result_Gratiot <- fit_phenology(data=data_Gratiot, 
		fitted.parameters=parg, fixed.parameters=NULL)
# Generate set of priors for Bayesian analysis
pmcmc <- phenology_MHmcmc_p(result_Gratiot, accept = TRUE)
# Here no WAIC data
result_Gratiot_mcmc <- phenology_MHmcmc(result = result_Gratiot, n.iter = 10000, 
                                        parametersMCMC = pmcmc, n.chains = 1, 
                                        n.adapt = 0, thin = 10, trace = FALSE, 
                                        WAIC=FALSE)
# Here the WAIC is at the level of the beaches
result_Gratiot_mcmc <- phenology_MHmcmc(result = result_Gratiot, n.iter = 10000, 
                                        parametersMCMC = pmcmc, n.chains = 1, 
                                        n.adapt = 0, thin = 10, trace = FALSE, 
                                        WAIC=TRUE, WAIC.bybeach=TRUE)
# Here the WAIC is at the level of the counts
result_Gratiot_mcmc <- phenology_MHmcmc(result = result_Gratiot, n.iter = 10000, 
                                        parametersMCMC = pmcmc, n.chains = 1, 
                                        n.adapt = 0, thin = 10, trace = FALSE, 
                                        WAIC=TRUE, WAIC.bybeach=FALSE)
loo::loo(result_Gratiot_mcmc$WAIC)
loo::waic(result_Gratiot_mcmc$WAIC)
# Get standard error of parameters
summary(result_Gratiot_mcmc)
# Make diagnostics of the mcmc results using coda package
mcmc <- as.mcmc(result_Gratiot_mcmc)
require(coda)
heidel.diag(mcmc)
raftery.diag(mcmc)
autocorr.diag(mcmc)
acf(mcmc[[1]][,"LengthB"], lag.max=200, bty="n", las=1)
acf(mcmc[[1]][,"Max_Complete"], lag.max=50, bty="n", las=1)
batchSE(mcmc, batchSize=100)
# The batch standard error procedure is usually thought to 
# be not as accurate as the time series methods used in summary
summary(mcmc)$statistics[,"Time-series SE"]
plot(result_Gratiot_mcmc, parameters="Max_Complete", las=1, xlim=c(-10, 210))
plot(result_Gratiot_mcmc, what="MarkovChain", ylim=c(20, 40))
plot(result_Gratiot_mcmc, what="LnL")
}