Internal helpers for computing the root mean squared error (RMSE) between predicted and observed quantiles and conditional accuracy functions.
calc_rmse_eval(pdfs, t_vec, dt, stats_agg, stats_agg_info, weight_err = 1.5)calc_rmse(quants_pred, cafs_pred, quants_obs, cafs_obs, weight_err = 1.5)
A single numeric RMSE value, or NULL if no observed stats were
provided, or Inf if predictions failed (contain NA).
list of PDFs per condition (named).
numeric time vector.
numeric time step.
list of observed summary statistics.
list with info on quantile probabilities and CAF bins.
non-negative numeric scalar; weight factor for CAF error relative to quantile error. Default is 1.5
numeric vector of predicted quantiles (already flattened).
numeric vector of predicted CAFs (already flattened).
numeric vector of observed quantiles (already flattened).
numeric vector of observed CAFs (already flattened).
calc_rmse_eval() prepares observed and predicted quantiles/CAFs from PDFs
and aggregated info, then calls calc_rmse().
calc_rmse() computes the weighted RMSE given predicted and observed
quantiles/CAFs.
stats_from_pdfs_agg_info()