Scott Spencer lectures in the applied analytics graduate program at Columbia University, his alma mater. He works with core developers of Stan, a probabilistic programming language, building Bayesian, generative models of complex processes involving both human behavior and physical phenomena.
He focuses modeling work on data for good, and on professional sports. Recent efforts are on the spatial and temporal impact of sea-level rise on the perceived value of coastal properties, and on major league baseball player and team performance. His analyses have extended to other domains as well. Previously, he analyzed data and communicated statistical insights, persuading various stakeholders of his technology clients (including Vevo, Freewheel, Johnson & Johnson, Dow Agrosciences, Qualcomm, Fitbod, and Amazon) on the impact of his analyses for decisions related to patent litigation and licensing.
To assist visual communication, he has developed multiple R packages, including one for mapping data values to visual encodings that are perceptually-uniform separately across hue, saturation, and luminance. The most persuasive communications are transparent and account for uncertainty, which are two areas of interest in his research and work in modeling, and quantitative communication through visualization and storytelling.
Along with honors recognition for research and writing, his visualizations have been showcased and longlisted in the Kantar Information is Beautiful Awards, and his modeling and analyses have won analytics competitions, including the Society for American Baseball Research’s analytics competition, graduate division for his work involving human decision-making based on perception of spatial and temporal event information.
- J.D., University of Illinois at Urbana-Champaign
- M.S., Columbia University
- B.S., Texas Tech University