probabilityDistributions

#install.packages("devtools")
#devtools::install_github("PRIMER-e/probabilityDistributions")
#devtools::install_github("PRIMER-e/responseFunctions")

library(probabilityDistributions)
library(responseFunctions)
library(rstan)
#> Loading required package: StanHeaders
#> Loading required package: ggplot2
#> rstan (Version 2.21.7, GitRev: 2e1f913d3ca3)
#> For execution on a local, multicore CPU with excess RAM we recommend calling
#> options(mc.cores = parallel::detectCores()).
#> To avoid recompilation of unchanged Stan programs, we recommend calling
#> rstan_options(auto_write = TRUE)

model_code <- paste( 
"
functions {
",
modskurt_stan[["source_code"]],
modskurtp1_stan[["source_code"]],
zi_neg_binomial_log_linked_2_lpmf_stan[["source_code"]],
"
}

data {
  int<lower=1> N;
  real x[N];
  int<lower=0> y[N];
}

// The parameters accepted by the model.
parameters {
real<lower=0> H;
real m;
real<lower=0> s;
real r;
real d;
real<lower=0> phi;
real gamma0;
real gamma1;
}

// The model to be estimated.
model {
// Having no prior defined is equivalent to all parameters having
// a uniform prior, which is exactly what we want for finding
// Maximum Likelihood (ML) estimates.

for (n in 1:N) {
  real mu = modskurtp1(x[n], H, m, s, r, d);
  target += zi_neg_binomial_log_linked_2_lpmf(y[n] | mu, phi, gamma0, gamma1);
  }
}
",
sep = "\n")

model <- stanc(model_code = model_code)