The Stan source code of the fitted model, with comments stripped for a clean
read. The raw source (comments included) is stored on the fit in
fit$meta$stancode.
Value
A kb_stancode object: the comment-stripped Stan source string, which
prints as readable code. Use as.character() for the plain string.
See also
Other generics:
augment.kb_fit(),
coef.kb_fit(),
converged.kb_fit(),
fitted.kb_fit_weight(),
glance.kb_fit(),
log_lik.kb_fit(),
posterior_epred.kb_fit_weight(),
posterior_linpred.kb_fit_weight(),
posterior_predict.kb_fit_weight(),
predict.kb_fit_weight(),
prior_summary.kb_fit(),
residuals.kb_fit_weight(),
samples(),
summary.kb_fit(),
tidy.kb_fit_weight()
Examples
kb_stancode(fit_weight_sim_nereo)
#> data {
#> int<lower=0> nObs;
#> int<lower=1> nSite;
#> int<lower=1> nYear;
#> array[nObs] int<lower=1, upper=nSite> site;
#> array[nObs] int<lower=1, upper=nYear> year;
#> vector<lower=0>[nObs] diameter;
#> vector<lower=0>[nObs] weight;
#> real<lower=0> diameter_ref;
#> real prior_intercept_mu;
#> real<lower=0> prior_intercept_sd;
#> real prior_diameter_mu;
#> real<lower=0> prior_diameter_sd;
#> real prior_diameter2_mu;
#> real<lower=0> prior_diameter2_sd;
#> real<lower=0> prior_sd_site_rate;
#> real<lower=0> prior_sd_site_diameter_rate;
#> real<lower=0> prior_sd_site_year_rate;
#> real<lower=0> prior_sd_residual_rate;
#> int<lower=0, upper=1> prior_only;
#> int<lower=0, upper=1> site_year_on;
#> }
#> transformed data {
#> real nu = 4.0;
#> real log_diameter_ref = log(diameter_ref);
#> vector[nObs] log_diameter = log(diameter) - log_diameter_ref;
#> vector[nObs] log_diameter_sq = log_diameter .* log_diameter;
#> vector[nObs] log_weight = log(weight);
#> array[nObs] int sy_idx;
#> for (i in 1:nObs) {
#> sy_idx[i] = site[i] + (year[i] - 1) * nSite;
#> }
#> }
#> parameters {
#> real bWeight;
#> real bDiameter;
#> real bDiameter2;
#> real<lower=0> sSite;
#> real<lower=0> sSiteDiameter;
#> real<lower=0> sSiteYear;
#> real<lower=0> sWeight;
#> vector[nSite] z_bSite;
#> vector[nSite] z_bSiteDiameter;
#> matrix[nSite, nYear] z_bSiteYear;
#> }
#> transformed parameters {
#> vector[nSite] bSite = z_bSite * sSite;
#> vector[nSite] bSiteDiameter = z_bSiteDiameter * sSiteDiameter;
#> matrix[nSite, nYear] bSiteYear = z_bSiteYear * sSiteYear;
#> vector[nObs] log_eWeight = bWeight + bSite[site]
#> + (bDiameter + bSiteDiameter[site]) .* log_diameter
#> + bDiameter2 * log_diameter_sq
#> + site_year_on * to_vector(bSiteYear)[sy_idx];
#> }
#> model {
#> bWeight ~ normal(prior_intercept_mu, prior_intercept_sd);
#> bDiameter ~ normal(prior_diameter_mu, prior_diameter_sd);
#> bDiameter2 ~ normal(prior_diameter2_mu, prior_diameter2_sd);
#> sSite ~ exponential(prior_sd_site_rate);
#> sSiteDiameter ~ exponential(prior_sd_site_diameter_rate);
#> sSiteYear ~ exponential(prior_sd_site_year_rate);
#> sWeight ~ exponential(prior_sd_residual_rate);
#> z_bSite ~ std_normal();
#> z_bSiteDiameter ~ std_normal();
#> to_vector(z_bSiteYear) ~ std_normal();
#> if (prior_only == 0) {
#> log_weight ~ student_t(nu, log_eWeight, sWeight);
#> }
#> }
#> generated quantities {
#> vector[nObs] log_lik;
#> vector[nObs] yrep;
#> for (i in 1:nObs) {
#> log_lik[i] = student_t_lpdf(log_weight[i] | nu, log_eWeight[i], sWeight);
#> yrep[i] = exp(student_t_rng(nu, log_eWeight[i], sWeight));
#> }
#> }