Skip to contents

Simulate Data from a Compound Poisson-Normal Model

simulate_cpn_data.Rd

This function generates synthetic data based on a Compound Poisson-Normal (CPN) model. The number of events for each observation is drawn from a Poisson distribution, and the outcome is the sum of normally distributed values for each event.

Usage

simulate_cpn_data(n = 100, beta = c(0.5, -0.3, 0.7), mu = 1, sigma = 2)

Arguments

n

Integer. Number of observations to simulate. Default is 100.

beta

Numeric vector. Coefficients for the linear predictor, including intercept. Default is c(0.5, -0.3, 0.7). Must match the number of columns in the model matrix: intercept, x1B, x2.

mu

Numeric. Mean of the Normal distribution for each event. Default is 1.

sigma

Numeric. Standard deviation of the Normal distribution for each event. Default is 2.

Value

A data.frame with n rows and 3 columns:

y

Numeric response variable generated from the CPN model.

x1

Categorical predictor with levels "A" and "B".

x2

Continuous predictor drawn from a standard normal distribution.

Details

The predictors include a binary categorical variable (x1) and a continuous variable (x2). A linear model with coefficients beta is used to model the log of the Poisson rate.

Examples

set.seed(42)
simulated_data <- simulate_cpn_data(n = 200)
head(simulated_data)
#>            y x1          x2
#> 1  0.0000000  A -0.71040656
#> 2 -2.4874819  A  0.25688371
#> 3  0.3583388  A -0.24669188
#> 4  0.0000000  B -0.34754260
#> 5  0.0000000  A -0.95161857
#> 6 16.0883605  B -0.04502772