Research using R underpins a wide array of applied decision-making. Growing recognition of irreproducibility of results produced using R raises concerns about the credibility of research and the reliability of decisions they inform.
Although there are many aspects to reproducible research, one major problem is the basic lack of computational reproducibility of published work – or the ability to rerun an analysis and reproduce the same results. There are many strategies for improving computational reproducibility, such as clear code structure, functional programming, and version control. Besides these tactics, however, we often find that analyses that reproduce on one machine or in one environment do not run in another, or do not run in the future. This problem is related to changing computing environments and software versions, and can be addressed using containerisation.
In this talk, Saras will introduce containerisation and its implementation using an open-access R package. This introduction should be accessible to those who have never heard of or used containerisation before, as well as those who already actively use services such as Docker for containerisation.
Saras Windecker is an ecologist and postdoctoral fellow with the Quantitative and Applied Ecology Group at the University of Melbourne. Check out her work on github at https://github.com/smwindecker