Calculate Model Posterior Distributions from Quantiles

Fits theoretical distributions to posterior quantiles and creates a posteriors.json file that mirrors the structure of the priors.json file with updated parameter values.

Usage

calc_model_posterior_distributions(
  quantiles_file = "./results/posterior_quantiles.csv",
  priors_file,
  output_dir = "./results",
  control = NULL,
  verbose = TRUE
)

Arguments

quantiles_file

Path to the posterior_quantiles.csv file (default: "./results/posterior_quantiles.csv")

priors_file

Path to the priors.json file to use as template

output_dir

Directory to save posteriors.json (default: "./results")

control

Optional control list (or path to control.json). When provided, sampling flags are used to distinguish frozen parameters (sampling flag FALSE, tagged "frozen") from genuinely converged parameters (sampling flag TRUE, tagged "fixed").

verbose

Logical; print progress messages (default: TRUE)

Value

List containing:

posteriors

The complete posteriors object

n_parameters_updated

Number of parameters successfully updated

n_parameters_failed

Number of parameters that failed to fit

output_file

Path to the created posteriors.json file

Details

The function:

1. Reads the posterior_quantiles.csv file produced by calc_model_posterior_quantiles

2. Processes all quantile rows in the file (user controls what to include)

3. Loads the priors.json as a template structure

4. For each posterior parameter in the quantiles table:

   • Extracts the 2.5%, 50%, and 97.5% quantiles

   • Calls the appropriate fit_*_from_ci function based on distribution type

   • Updates the corresponding entry in the posteriors structure

5. Removes from the output every parameter (global, or location/parameter pair) that does not appear in the quantiles file, so posteriors.json holds only the estimated parameters and is not a drop-in full priors file (use update_priors_from_posteriors to merge it back into a priors object). Rows of scale "unknown" (columns outside the estimated_parameters template) are skipped silently and are not counted as failures

6. Properly handles location-specific parameters for single or multiple countries

7. Writes posteriors.json to the output directory

Examples

if (FALSE) { # \dontrun{
# Standard usage after running calc_model_posterior_quantiles
posterior_dists <- calc_model_posterior_distributions(
  quantiles_file = "./results/posterior_quantiles.csv",
  priors_file = "./config/priors.json",
  output_dir = "./results"
)
} # }