generate_model_data {bennu}

generate model data for testing purposes

Description

Simulate kits ordered and kits distributed for a set number of regions and time-points.

The kits ordered simulation is a simple square-term multiplied by region_coeffs. For example if region_coeffs = c(1,2) then the number of kits ordered at month 12 are c(1,2) * 12^2 = c(144,288).

The probability of kit use in time is assumed to increase linearly in inverse logit space at a constant rate 0.1. The probability of reporting for each month and region is iid distributed logit^{-1}(p) \sim N(2,5) which produces a mean reporting rate of approximately 88%

Usage

generate_model_data(
  N_t = 24,
  region_coeffs = c(5, 0.5),
  c_region = c(-1, 2),
  reporting_freq = NULL
)

Arguments

N_t

number of time-points

region_coeffs

vector of coefficients for regions determining kit orders

c_region

logit probability of kit use per region

reporting_freq

The frequency that distribution data is provided. If NULL distribution frequency matches orders frequency

Value

A tibble

Orders

Kit orders per time and region

regions

Numeric index indicating region of orders and distributions

Reported_Used

Number of kits reported as used

Reported_Distributed

Number of kits reported as distributed

p_use

Probability that a kit was used

p_reported

Probability that a distributed kit was reported

times

Index for time

region_name

String index for the region

See Also

Other data generation: model_random_walk_data()