Summarize shared curve-parameter posteriors — summarize_curve_params

Extracts the quantities that Chapter 1 (model.jags) and Model 2a (model_2a.jags) have in common so they can be compared directly: the population means mu.par[k, p] and the within-biomarker variances. For Model 2a the within-biomarker variance is the marginal variance diag(solve(prec.par[k])) + lambda[k, ]^2 (set with_loadings = TRUE); for Chapter 1 there are no loadings, so it is just diag(solve(prec.par[k])) (with_loadings = FALSE). Posterior medians are returned.

Pure extraction/summary (delegates the variance algebra to marginal_var_2a()); no fitting.

Usage

summarize_curve_params_2a(mcmc, with_loadings = FALSE, param_names = NULL)

Arguments

mcmc: A coda::mcmc.list (or a named draws matrix) containing monitored mu.par and prec.par (and lambda when with_loadings).
with_loadings: logical; add the squared factor loadings to the within-biomarker variance (TRUE for Model 2a, FALSE for Chapter 1).
param_names: Optional length-P parameter labels (defaults to the log-scale names).

Value

A data.frame with columns biomarker (index), param, mean_med (median of mu.par), and var_med (median within-biomarker variance).