Generate artificial delay data. — .generate_delay

This utility can be used to build toy examples to test functions dealing with empirical delay data. It is very basic in what it simulates. A random walk is simulated over n_time_steps, representing the incidence through time. The result of this simulation is offset so that all values are positive. Then, for each time step, n samples from a delay distribution are taken, with n being the incidence value at this time step. The random draws are then multiplied by a factor (>1 or <1) to simulate a gradual shift in the delay distribution through time. This multiplication factor is calculated by linearly interpolating between 1 (at the first time step), and delay_ratio_start_to_end linearly, from 1 at the first time step to ratio_delay_end_to_start at the last time step.

.generate_delay_data(
  origin_date = as.Date("2020-02-01"),
  n_time_steps = 100,
  time_step = "day",
  ratio_delay_end_to_start = 2,
  distribution_initial_delay = list(name = "gamma", shape = 6, scale = 5),
  seed = NULL
)

Arguments

origin_date	Date of first infection.
n_time_steps	interger. Number of time steps to generate delays over
time_step	string. Time between two consecutive incidence datapoints. "day", "2 days", "week", "year"... (see `seq.Date` for details)
ratio_delay_end_to_start	numeric value. Shift in delay distribution from start to end.
distribution_initial_delay	Distribution in list format.
seed	integer. Optional RNG seed.

Value

dataframe. Simulated delay data.