Title: When a Dashboard Won’t Do: Automating Static Data Visualization of COVID-19 Data in Python

Abstract:

When a project demands ongoing updates to data visualizations, the go-to solution is often to build a dashboard: an application that connects to a database and updates visualizations in real time as data changes. However, in many projects, such an approach is not feasible or sufficient. Tight timelines, security concerns, and computing constraints are common culprits. In such situations, analysts often must generate routine reports the old-fashioned way: as static files such as PowerPoints and PDFs. How can we strive toward the goals of automation and replicability when generating these reports?

In this presentation, we discuss an approach to automating the generation of weekly PowerPoint reports containing data visualizations. The reports present weekly COVID-19 model output to state government stakeholders. We rely on the Python packages matplotlib and python-pptx. From the user’s perspective, we replicate the look and feel of an interactive dashboard as much as possible. From the developer’s perspective, generating these reports requires no more ongoing time commitment than the maintenance of a dashboard. This approach is efficient, practical, and generalizable to any situation where time can be saved by automating the creation of routine reports.