Abstract: 

As statistical and machine learning techniques become part of nearly all data-driven organizations, practitioners have a responsibility to communicate about the assumptions and mechanics of their approaches (not just their conclusions). The abundance of statistical packages in languages such as R and Python have made these analysis strategies easier than ever to implement. However, there has been comparably little development in the platforms for communicating the assumptions, limitations, and intentions of these models to broader audiences. This communication gap greatly diminishes the usability of pertinent data insights.

This talk will describe a process for designing and building visual representations of statistical and machine learning concepts. Drawing upon keynote examples of visual explanations of analytical techniques, this talk will introduce a set of steps for visually communicating complex topics. By isolating specific concepts and mapping them to a data structure, analysts are able to visualize the underlying concepts of interest. Drawing on visualization theory, participants will learn optimal processes for selecting visual encodings to express analytical concepts. Whether you are using D3.js or a whiteboard, identifying a data structure that represents a concept can then enable you to express ideas to your specific audience.

In order to make accurate and responsible data-driven decisions, decision makers need to have a more robust understanding of the analytical processes applied to their data. We, as practitioners, are responsible for creating resources for amplifying the understanding of the individuals that consume our data. Visualization provides a powerful tool not just for expressing our data, but for expressing our methods as well.

Bio: 

Michael Freeman is a faculty member at the University of Washington's Information School where he teaches courses in Data Science, Interactive Data Visualization, and Web Development. Prior to his teaching career, he levered D3.js as a Data Visualization Specialist at the Institute for Health Metrics and Evaluation. There, he built a variety of interactive visualization systems to help researchers and the public explore global health trends. Michael is interested in applications of data visualization to social change, and holds a Master's in Public Health from the University of Washington.