Software

I like to build tools that help researchers. Below are two examples of tools that I have worked on that are actively used within the research community.

Bayesplay

Bayesplay was initially created as an R package for exploring and understanding Bayesian analyses and for computing Bayes factors for simple 1-parameter models. The motivation for Bayesplay was the desire to make these kinds of analyses more transparent and understandable. In creating it, I was explicitly aiming for an approach that would sit somewhere between easy-to-use, but opaque, packages such as bayesfactor, and probabilistic programming languages such as stan.

Bayesplay makes Bayes factor analyses transparent by making the steps in the model comparison explicit. Researchers have to describe the data with a likelihood, specify each model’s prior, and compare how well each model predicts the observations.

Since its initial release as an R package, I have also released a web app that makes it possible to interactively specify models, calculate Bayes factors, and generate the corresponding R code. Bayesplay also powers the General Bayesian Tests module in JASP.

Bayesplay is actively used in the research community, including by researchers in papers published by researchers at the University of Cambridge, the University of Sussex, the University of Amsterdam, and the Massachusetts Institute of Technology.

Try the web app · Read the documentation · Browse the source

CATS: the Climate-Aware Task Scheduler

CATS is helping UK researchers schedule compute jobs to take advantage of low-carbon electricity. It works with existing job schedulers, including at and Slurm’s sbatch.

To learn more about CATS, including its command-line documentation, see our recent paper published in the Journal of Open-Source Software.

Read the documentation · Browse the source · Read the paper

For related research, see my publications.