Apply a nowcast calibration to quantile predictions

Rescales each quantile by moving it away from (or toward) the median by the learned per-group `factor`. By construction the rescaled central interval covers `level` of the BACKTEST the factor was learned on; that is not a guarantee about future weeks, and the median is left unchanged. Groups with no learned factor (e.g. an unseen horizon) pass through unchanged.

Usage

nowcast_apply_calibration_v1(x, calibration)

Arguments

x

Long quantile predictions (`reference`, the calibration's `by` column(s), `quantile_level`, `predicted`) – e.g. a fresh [nowcast_backtest] output or a melted collapse.

calibration

A `nowcast_calibration` from [nowcast_estimate_calibration_v1].

Value

`x` with `predicted` recalibrated.

See also

Neither package vignette covers calibration. The example below is its only worked demonstration.

Other nowcast calibration functions: nowcast_estimate_calibration_v1(), print.nowcast_calibration()

Examples

w <- cstime::dates_by_isoyearweek$isoyearweek
i <- match("2023-01", w)
set.seed(1)
d <- data.table::data.table(
  isoyearweek_reference = w[i + rep(0:39, each = 3)],
  isoyearweek_reporting = w[i + rep(0:39, each = 3) + rep(0:2, 40)],
  numerator = rpois(120, c(30, 15, 5)),
  indicator_tag = "x", location_code = "nation", age = "total", sex = "total"
)
d <- d[isoyearweek_reporting <= w[i + 39]]
tri <- csfmt_reporting_triangle_v3(
  d,
  id_cols = c("indicator_tag", "location_code", "age", "sex")
)

method <- function(x) nowcast_quasipoisson_v1(x, max_delay = 3, n_sim = 200)
bt <- nowcast_backtest(
  tri, method,
  max_delay = 3, as_of_weeks = w[i + 20:38], horizons = 0:1,
  probs = c(0.05, 0.5, 0.95), seed = 1
)
cal <- nowcast_estimate_calibration_v1(bt, nowcast_truth(tri, max_delay = 3))

adj <- nowcast_apply_calibration_v1(bt, cal)

# the median is untouched; the other two quantiles move toward or away from
# it, so the interval width changes by the learned factor
width <- function(x) {
  x[horizon == 0, .(width = diff(range(predicted))), by = reference][1:3]
}
width(bt)
#>    reference width
#>       <char> <num>
#> 1:   2023-21 23.00
#> 2:   2023-22 23.05
#> 3:   2023-23 24.05
width(adj)
#>    reference    width
#>       <char>    <num>
#> 1:   2023-21 13.59300
#> 2:   2023-22 13.62255
#> 3:   2023-23 14.21355