Functions

convert(::Type{DataFrame}, pdata::Vector{Pseudodata})

Convert the vector of Pseudodata type to a DataFrame

logpdf(pdata::Pseudodata, y::Vector{<:Real})

Compute the log density of each of the potential data y according to the distributions in pdata.

logpdf(pensemble::Pseudoensemble, y::Vector{<:Real})

Compute the logpdf of each of the potential data y according to the distributions in pensemble.

Details

Independance is assumed between the members of the pseudoensemble.value, i.e. the sum of the logpdf of each member is taken.

pdf(pdata::Pseudodata, y::Vector{<:Real})

Compute the density of each of the potential data y according to the distributions in pdata.

ensemblemean(pdata::Vector{Pseudodata})

Compute the ensemble mean for each year.

convert(::Type{MaximumLikelihoodEVA}, fm::PseudoMaximaEVA, iter::Int)

Convert a single MCMC iteration of the fm model to a pseudo MaximumLikelihoodEVA for hacking Extremes.jl methods.

function logpdf(fm::PseudoMaximaEVA)

Compute the log density of fm for all MCMC iterations.

dic(fm::PseudoMaximaEVA)

Compute the Deviance Information Criterion (DIC) described by Gelman et al. (2013) for the PseudoMaximaEVA model fm.

Details

Reference: Gelman, A., Carlin, J.B., Stern, H.S., Dunson, D.B., Vehtari, A. & Rubin, D.B. (2013). Bayesian Data Analysis (3rd ed.). Chapman and Hall/CRC. https://doi.org/10.1201/b16018

quantile(fm::PseudoMaximaEVA, order::Real)

Compute the effective quantiles or order order for each MCMC iteration of model fm.

PseudoMaximaEVA(pseudodata::Pseudoensemble, model::EVA, sim::Mamba.Chains)

Construct a PseudoMaximaEVA type.

Details

Encapsulates the probability distributions for each of the unobserved data for an ensemble of pseudodata, the extreme value model and a sample from the parameters posterior distribution.

TODO: Verify if all member has the same year vector.

PseudoMaximaModel(data::Vector{Pseudodata};
    locationcov::Vector{Variable} = Vector{Variable}(),
    logscalecov::Vector{Variable} = Vector{Variable}(),
    shapecov::Vector{Variable} = Vector{Variable}())

Creates a PseudoMaximaModel structure.

Pseudodata(name::String, value::Vector{<:UnivariateDistribution})

Construct a Pseudodata type.

Details

Encapsulates the probability distributions for each of the unobserved data. The data are not directly observed but their distributions are known. The distributions can be different for each of the data.