Compress a csfmt_ensemble_v4 into a table to store

Returns $data of x as a data.table, with the median and 9 percentiles of each base measure in place of its draws. as_ensemble_v4() turns the table back into an ensemble.

Usage

ens_compress_v4(x)

Arguments

x

A csfmt_ensemble_v4, full or compressed. Each column of $data MUST be an atomic vector that is not a factor.

Value

A data.table with the columns of x$data and the columns of each base measure. Each column name MUST be at most 63 bytes, and the function stops with an error when one is not.

Details

A base measure is a draw matrix whose name does not end in a reserved suffix, such as _trend_beta1 or _hlm. Each base measure <m> becomes these columns:

• <m>: the median of the draws.

• <m>_q02x5 to <m>_q97x5: the draws at the 9 percentiles 2.5, 5, 10, 25, 50, 75, 90, 95 and 97.5.

The percentiles are NA in a row where every draw that is not NA has the same value. A row with one draw, as from nowcast_passthrough_v4(), is such a row. The table does not hold the derived draw matrices. The columns that the steps wrote to $data already summarise them.

The function draws no random numbers.

See also

Other v4 functions: as_ensemble_v4(), csfmt_ensemble_v4(), csfmt_reporting_triangle_v4(), ens_add_rate_v4(), mem_thresholds_v4(), nowcast_ecdf_v4(), nowcast_passthrough_v4(), print.csfmt_ensemble_v4(), short_term_trend_v4(), signal_detection_hlm_v4()

Examples

d <- data.table::data.table(
  location_code = "nation",
  isoyearweek = sprintf("2023-%02d", 1:10)
)
set.seed(1)
ens <- csfmt_ensemble_v4(
  d,
  id_cols = "location_code",
  draws = list(cases = matrix(rpois(10 * 200, 20), nrow = 10))
)
stored <- ens_compress_v4(ens)
stored[, .(isoyearweek, cases, cases_q02x5, cases_q97x5)]
#>     isoyearweek cases cases_q02x5 cases_q97x5
#>          <char> <num>       <num>       <num>
#>  1:     2023-01    20      11.975      29.000
#>  2:     2023-02    20      11.000      29.000
#>  3:     2023-03    20      12.000      29.000
#>  4:     2023-04    19      11.000      29.025
#>  5:     2023-05    19      11.975      29.000
#>  6:     2023-06    20      11.975      29.025
#>  7:     2023-07    20      11.975      27.025
#>  8:     2023-08    19      12.975      28.000
#>  9:     2023-09    20      11.000      29.000
#> 10:     2023-10    19      12.000      29.025