Tidy the delay distribution from an epidist fit

epidist is the one supported package that does not produce a nowcast. It estimates the reporting-delay distribution, so there are no per-event-date case estimates to tidy and the columns tidy.nowcast() promises (event_date, stratum, ...) would all be meaningless. This method therefore returns a different, delay-shaped table – one row per distribution parameter rather than one row per date.

Usage

# S3 method for class 'epidist_fit'
tidy(x, probs = NULL, level = 0.95, newdata = NULL, ...)

Arguments

x

A fit from epidist::epidist().

probs

Optional numeric vector of probabilities in [0, 1], adding one q* column each.

level

Width of the reported interval. Defaults to 0.95.

newdata

Optional data frame passed to epidist::predict_delay_parameters(), for a fit with covariates in the delay model (formula = mu ~ 1 + gender, say). NULL uses the fit's own data.

...

Unused, for generic consistency.

Value

A tibble, as described in Value.

Value

A tibble with one row per parameter of the fitted delay distribution and these columns:

term

character. The parameter: the distribution's own parameters (mu, sigma, ... – whichever the family has) plus the derived mean and sd, which are the numbers most people actually want.

estimate

numeric. Posterior median.

conf.low, conf.high

numeric. Interval bounds, following broom's naming.

level

numeric. The width of that interval.

engine

character. Always "epidist".

probs appends one q* column per requested probability, exactly as it does for the nowcast methods – the fit exposes draws, so any quantile is real rather than an approximation.

Dispatch

epidist() returns an object of class c("brmsfit", "epidist_fit"), in that order, so if broom.mixed is loaded its tidy.brmsfit() method matches first and you get raw brms parameters instead of this table. Call tidy.epidist_fit(fit) explicitly when you want the delay distribution and cannot be sure which method will win.

See also

tidy.nowcast() for the case-count nowcast engines, tbl_now_to_epidist() for the conversion.

Examples

data(denguedat)
# A short window: fitting a delay distribution does not need twenty years of
# data, and Stan is slow.
recent <- subset(denguedat, onset_week >= as.Date("2010-06-01"))
nowobj <- tbl_now(recent,
  event_date = "onset_week", report_date = "report_week", verbose = FALSE
)

## The conversion itself is quick, and is what tidy() will later summarise.
converted <- suppressWarnings(tbl_now_to_epidist(nowobj, verbose = FALSE))
#> ℹ No observation time column provided, using 2010-12-27 as the observation date (the maximum of the secondary event upper bound).
head(converted)
#> # A tibble: 6 × 10
#>   ptime_lwr ptime_upr stime_lwr stime_upr obs_time pdate_lwr  pdate_upr 
#>       <dbl>     <dbl>     <dbl>     <dbl>    <dbl> <date>     <date>    
#> 1         0         7         7        14      203 2010-06-07 2010-06-14
#> 2         0         7         7        14      203 2010-06-07 2010-06-14
#> 3         0         7         7        14      203 2010-06-07 2010-06-14
#> 4         0         7         7        14      203 2010-06-07 2010-06-14
#> 5         0         7         7        14      203 2010-06-07 2010-06-14
#> 6         0         7         7        14      203 2010-06-07 2010-06-14
#> # ℹ 3 more variables: sdate_lwr <date>, sdate_upr <date>, obs_date <date>

# Fitting compiles a Stan model, so this takes about a minute even on a short
## chain. `try()` guards the case where \pkg{epidist} is installed but its Stan
# toolchain is not; use \pkg{brms}'s defaults for real work.
fit <- try(
  converted |>
    epidist::as_epidist_marginal_model() |>
    epidist::epidist(chains = 1, iter = 200, refresh = 0),
  silent = TRUE
)
#> ! Setting 560 observation times beyond 182 (=2x max delay) to Inf. This
#>   improves model efficiency by reducing unique observation times while
#>   maintaining model accuracy as these times should have negligible impact.
#> Warning: Found infinite values in the data, which may cause issues for Stan.
#> ℹ Data summarised by unique combinations of:
#> * Model variables: delay bounds, observation time, and primary censoring window
#> ! Reduced from 5426 to 116 rows.
#> ℹ This should improve model efficiency with no loss of information.
#> Warning: Found infinite values in the data, which may cause issues for Stan.
#> Warning: Found infinite values in the data, which may cause issues for Stan.
#> Compiling Stan program...

if (!inherits(fit, "try-error")) {
  print(tidy(fit))
  print(tidy(fit, probs = c(0.05, 0.95)))
}