Create summary statistics from fitted BFACT models across multiple H values. Combines individual fit objects into posterior summaries and model comparison metrics.
Arguments
- fits
List of fit objects (output from
fit_bfact_model()), or a list of such lists for multiple replications. See details.- sim
List containing simulation data (Y, z, years), or list of such lists for multiple replications.
- by_replicate
Logical. If TRUE, compute per-replicate statistics (default FALSE). If fits and sim are lists of lists, this will aggregate across replications.
Value
A list containing:
timewise_summaries: tibble with posterior means/CIs per H
model_comparison: tibble comparing metrics across H values
per_replicate_stats: per-replicate RMSE if by_replicate=TRUE
replicate_comparison: tibble with mean/sd RMSE by H across replications (if multiple reps)
Details
Consolidates multiple H fits into a single summary table with columns:
H_fit: fitted H value
time: time index
post_mean: posterior mean prediction
post_ci_lower, post_ci_upper: 95% credible interval
se: squared error at each time point
rmse: overall RMSE for that H
For multiple replications, pass:
fits: list of lists, where each inner list is fits for one replicate
sim: list of simulation lists, one per replicate
Examples
if (FALSE) { # \dontrun{
sim <- simulate_data(H_true = 2, K = 20)
fit_H2 <- fit_bfact_model(sim, H = 2)
fit_H3 <- fit_bfact_model(sim, H = 3)
results <- consolidate_results(
fits = list(fit_H2, fit_H3),
sim = sim
)
head(results$timewise_summaries)
} # }