Skip to contents

Build the priors mu_jt block from annual CFR estimates

Source: R/make_mu_jt.R
dot-mosaic_mu_jt_prior.Rd

The prior the integrated deaths likelihood reads (priors$mu_jt): per location and year the centre (logit_mean, the value make_mu_jt() puts in the config) and the SE of the country-trend mean (logit_se), plus the global widths. Used by data-raw/make_priors_default.R and by run_rolling_cv() for its per-cutoff priors.

Usage

.mosaic_mu_jt_prior(
  cfr_estimates,
  location_name,
  sd_year,
  tau,
  sd_product = 0.3,
  year_min = 2010L
)

Arguments

cfr_estimates

Data frame with iso_code, year, logit_mean and cfr_se (est_CFR_hierarchical()$predictions).

location_name

Character vector of ISO codes to include.

sd_year

Positive scalar: the GAM country-year SD (sigma).

tau

Scalar: the GAM between-country SD (reference only).

sd_product

Positive scalar: residual error of the GAM centre against the observed reported CFR in a calibration window, on the logit scale.

year_min

Integer: first year kept.

Value

A list with description, sd_year, sd_product, tau and location (one list of year, logit_mean, logit_se per ISO code).