Joint endemic-epidemic model

HHH4() fits surveillance::hhh4() jointly to all series and returns simulated forecasts for get_cv() and get_fcast().

Usage

HHH4(formula, control, neighbourhood = NULL, population = NULL, n_sim = 500L)

Arguments

formula: Use observation. Units come from the single key supplied to check_data().
control: A surveillance::hhh4() control list.
neighbourhood: Optional named adjacency or neighbourhood-order matrix.
population: Optional named population vector or time-by-unit matrix.
n_sim: Number of joint forecast simulations. Defaults to 500.

Value

A joint model specification for get_cv() or get_fcast().

Details

Model components and control options are documented in surveillance::hhh4(). insight.cast manages control$subset, so do not supply it. Data must have one series key, and all unit-indexed inputs must be named with its values. Fractional observations are rounded to counts before fitting.

Author

Cyril Geismar

Examples

if (FALSE) { # \dontrun{
states <- c("CT", "NY")
W <- matrix(
  c(0, 1, 1, 0),
  2, 2,
  byrow = TRUE,
  dimnames = list(states, states)
)
population <- c(CT = 3.6, NY = 20)
population <- population / sum(population)

get_data("flu", tolower(states)) |>
  get_cv(
    h = 4,
    n_origins = 8,
    models = list(
      HHH4 = HHH4(
        observation,
        control = list(
          ar = list(f = ~1),
          ne = list(f = ~1, normalize = TRUE),
          end = list(
            f = surveillance::addSeason2formula(~1, S = 1, period = 52)
          ),
          family = "NegBin1"
        ),
        neighbourhood = W,
        population = population
      )
    )
  )
} # }