Get a probabilistic draw from the observation model

The function ingests an estimate of a point nowcast, in the form of a point nowcast matrix and a vector of dispersion parameters, and draws from a negative binomial to get a single expected observed probabilistic draw of a nowcast

Usage

get_nowcast_mat_draw(point_nowcast_matrix, disp)

Arguments

point_nowcast_matrix

Matrix of both observations and point estimates of counts at each reference time and delay.

disp

Vector of dispersion parameters indexed starting at delay = 1,

Value

nowcast_w_obs_error Matrix containing the same upper left values as the point_nowcast_matrix, with the bottom right containing probabilistic draws assuming a negative binomial observation model with a mean given by the lower right entries in point_nowcast_matrix and dispersion parameters from disp

Examples

point_nowcast_matrix <- matrix(
  c(
    80, 50, 25, 10,
    100, 50, 30, 20,
    90, 45, 25, 17.8,
    80, 40, 23.2, 15.9,
    70, 35, 20.3, 19.9
  ),
  nrow = 5,
  byrow = TRUE
)

exp_obs_nowcast <- get_nowcast_mat_draw(
  point_nowcast_matrix = point_nowcast_matrix,
  disp = c(8, 1.4, 4)
)
print(exp_obs_nowcast)
#>      [,1] [,2] [,3] [,4]
#> [1,]   80   50   25   10
#> [2,]  100   50   30   20
#> [3,]   90   45   25   14
#> [4,]   80   40    0   23
#> [5,]   70   38   35    8