The Challenge:
To create a template workflow for a stream of analytics and data presentation projects for research and operational use at GOSH. To ensure that the codebase is: interpretable, reusable & maintainable and to control technical debt around the solutions we provide, so that they can be handed over to other technical teams in the Trust.
The Solution:
We have created a ‘Data Science’ template, inspired by Cookiecutter Data Science and using the targets R package to logically break down code into small reusable chunks applied to our research datasets.
The Results:
We are now running (and have completed) a number of projects from estimation of surgery lengths with XGBOOST, to dashboarding cardiac data for MDTs, to creating articles on historic data trends at GOSH and many more. This has provided us a rich common codebase applicable in all sorts of contexts, as well as giving us a headstart when understanding each other’s code and collaborating on development.