Dylan Halpern is the Principal Software Engineer for the US Covid Atlas at the Center. Utilizing methods of geospatial data analytics, visualization, and web development, he works in domains of public health, urban experience and activity, and transit. He holds a Master in City Planning from MIT.
In response to the COVID-19 pandemic, a research coalition led by the Center for Spatial Data Science at UChicago developed a first-of-its-kind dashboard that enables client-side analytics to perform spatial autocorrelation methods (LISA / Local Moran’s I) and binning algorithms including jenks natural breaks and hinge breaks directly in the client-side environment. The advances open up new possibilities for inexpensively hosted geospatial web applications with more robust analytic capacities than previously possible, all while utilizing free or low-cost web hosting services. Additionally, new visualization methods utilizing powerful WebGL-based geospatial libraries (Deck.gl, Kepler.gl) allow for more complex on-the-fly visualizations such as point-grid layers and dot density maps with robust filtering options.
This talk outlines the infrastructure used in the US Covid Atlas project and the capacities of jsgeoda (Xun Li, CSDS) to perform a variety of clustering and binning algorithms in the browser. Code demonstrations for main thread and worker implementations will be presented, along with a series of template Github repositories available for use licensed under GPL. By the end of this talk, viewers will be able to understand and replicate the US Covid Atlas’s web infrastructure, implement a simple jsgeoda instance for spatial clustering, and build from open source templates for their own geospatial analytics web application that does not require backend services.