Training workshop on drjacoby for Bayesian model fitting in R

This repository holds the material for a training workshop on drjacoby.

What to do before the workshop

1. Install drjacoby

Please install drjacoby in advance so we can hit the ground running. This will also help identify any issues with installation so we can iron them out in a less pressured environment!

You should be able to install directly from the R-universe using the following code. If it asks you to install from sources, type "Y" for Yes

install.packages('drjacoby', repos = 'https://mrc-ide.r-universe.dev')

The R-universe method is quite new, and so may not work. If it fails then please follow these instructions instead.

This repository assumes you are using drjacoby version 1.5.4. If you use older or newer versions, there is no guarantee that all functions will work as described here.

2. Download this repository

You will need a local copy of all the material in this repository. There are a couple of ways to get it. First, you could clone the repository. Another option would be to click on the Code button above and download the whole lot as a zip.

What the workshop will cover

We will cover the following topics:

Getting drjacoby up and running with the correct input formats
Producing basic plots of MCMC output and checking MCMC diagnostics
Model fit and model comparison. Credible intervals and posterior predictive checks
Activating parallel tempering to improve mixing and deal with multi-modality
Linking with other packages. Example of a compartmental model in Odin
Using C++ functions and running in parallel to speed things up

Structure of the material

All the lectures are contained in the Lectures folder, and each lecture has a corresponding numbered script in the R_scripts folder.

The main activity of the workshop will be to work through the scripts in the R_scripts folder. These are heavily commented with instructions and challenges directly in the script. You are encouraged (actually required) to edit these scripts, so please get used to modifying them! You should work from the drjacoby_training.Rproj to ensure all file paths work correctly.