To recap Tuesday’s sessions on “Storytelling with Data and Code,” the Research Data Management team here at Data Services offered a workshop on systems to help integrate code with narrative, and in particular the highly versatile Jupyter Notebooks and R Markdown approaches. These packages offer researchers a way to implement the increasingly popular “executable paper” whereby scripts that perform data transformations and analysis can be intermingled with important description and narrative in a document that can easily be moved to the web.
We started by looking at a few examples of such notebook publications and considering the most effective ways to group code and outline narrative to effectively communicate with readers. We went over good approaches to organizing the underlying projects files that do the work behind the executable paper. We then reviewed the features of Jupyter Notebooks (for Python, R, and Julia users) and R Markdown (for users of R) and did some hands-on practice with their interfaces to see how each approached organizing code. For Jupyter, there were some helpful quick start commands that were important, and we looked at two ways to move a notebook to the web. R Markdown provides a similar means of implementing an executable paper on the web, though with a slightly different approach to hosting a finished R Markdown file on the web.
Tuesday’s session continued with a talk by data journalist Meredith Broussard on key principles behind making visualizations for readers. Focusing on how to make numerical discussions clear to readers who often struggle to understand statistical descriptions, Broussard noted the importance of knowing one’s audience and tailoring the visualization to underscore the message (“edit for the angle”).
We’ll be continuing to integrate skills on these and other data + code storytelling platforms, so check back regularly on the class schedule at Data Services!