ZombieSim

The goal of the ZombieSim package is to provide tools for inference for models with intractable likelihoods using Approximate Bayesian Computation (ABC). It uses R with integrated C++ code (using Rcpp and RcppArmadillo) to help with computational speed.

This also contains functions to simulate a zombie epidemic under the model described in "When zombies attack!: Mathematical modelling of an outbreak of zombie infection". We use the SZR model given in the paper with the natural birth and death rate set to zero.

Installation

This package can be downloaded and used with devtools as below:

# install.packages("devtools")
library(devtools)
devtools::install_github("RaspberryEmma/ZombieSim")
library(ZombieSim)

Example

Lets now run through an example of how to use various parameters from the package:

# Let's simulate some data!
N <-5000
initial <- generateStartCond(N, initial.inf = 10)
new.example <- generateSZRdata(initial, total.T = 50)

# Let's plot our data!
plot_SZR(new.example$results)

# Let's get our samples for the posterior of the parameters
simulated.mat <- new.example$results
simulated.mat  <- as.matrix(simulated.mat[,2:4])
head(simulated.mat )
priorMin <- c(0.000004,0.000004,  0.000004) 
priorMax <- c(0.0015 ,0.001, 0.001)
res <- abcRej(simulated.mat , 10000000, 1000, priorMin, priorMax)

# Let's plot our simulated values alongside the true parameters!
library(ggplot2)
library(dplyr)
means <- c(new.example$beta, new.example$kappa, new.example$rho)
samples <- res %>%
  as_tibble()
colnames(samples) <- c("beta", "kappa", "rho")  
plots <- plot.posterior.samples(samples, means)
plots[[1]]

Have fun using the package! Any issues with the code, please get in contact!