cocoSim function

Simulation of Count Time Series

Simulation of Count Time Series

The function generates a time series of low counts from the (G)PAR model class for a specified innovation distribution, sample size, lag order, and parameter values.

cocoSim( type, order, par, length, xreg = NULL, init = NULL, julia = FALSE, julia_seed = NULL )

Arguments

  • type: character, either "Poisson" or "GP" indicating the type of the innovation distribution
  • order: integer, either 1 or 2 indicating the order of the model
  • par: numeric vector, the parameters of the model, the number of elements in the vector depends on the type and order specified.
  • length: integer, the number of observations in the generated time series
  • xreg: data.frame, data frame of control variables
  • init: numeric vector, initial data to use, default is NULL. See details for more information on the usage.
  • julia: If TRUE, the Julia implementation is used. In this case, init is ignored but it might be faster.
  • julia_seed: Seed for the Julia implementation. Only used if Julia equals TRUE.

Returns

a vector of the simulated time series.

Details

The function checks for valid input of the type, order, parameters, and initial data before generating the time series.

The init parameter allows users to set a custom burn-in period for the simulation. By default, when simulating with covariates, no burn-in period is specified since there is no clear choice on the covariates. However, the init argument gives users the flexibility to select an appropriate burn-in period for the covariate case. One way to do this is to simulate a time series using cocoSim with appropriate covariates and pass the resulting time series to the init argument of a new cocoSim run so that the first time series is used as the burn-in period. If init is not specified for the covariate case, a warning will be returned to prompt the user to specify a custom burn-in period. This helps ensure that the simulation accurately captures the dynamics of the system being modeled.

Examples

lambda <- 1 alpha <- 0.4 set.seed(12345) # Simulate using the RCPP implementation data_rcpp <- cocoSim(order = 1, type = "Poisson", par = c(lambda, alpha), length = 100) # Simulate using the Julia implementation data_julia <- cocoSim(order = 1, type = "Poisson", par = c(lambda, alpha), length = 100)

Author(s)

Manuel Huth

  • Maintainer: Manuel Huth
  • License: MIT + file LICENSE
  • Last published: 2023-10-01

Useful links