Online Program

Return to main conference page

All Times ET

Friday, June 4
Data Visualization
Honoring the Data Science Accomplishments of William S. Cleveland and John W. Tukey
Fri, Jun 4, 3:20 PM - 4:55 PM
TBD
 

How John Tukey Led Us Into Data Science (309873)

*William S. Cleveland, Purdue University 

Keywords: Visualization, Computational Environments, Computer Languages, Analytic Methods, Models

Tukey did it with brilliance, vision, innovation, and engagement. He injected into statistics many analytic methods. Some were classically statistical with probabilities and inference, and others defined algorithmically, pioneering machine learning before it had a name. He invented tools for numerical spectrum analysis, including the fast Fourier transform, which became widely used in many fields. He advocated exploratory data analysis and data visualization and wrote a book about it. He led the development of dynamic visualization, heading development of PRIM-9 for exploring multidimensional data using by point-cloud rotation controlled by ``projection pursuit''. Tukey was 1/2 time an executive director at Bell Labs and 1/2 time a professor in math at Princeton. He emphasized strongly that computing was critical for all technical fields. This led Bell Labs technical staff to develop UNIX, C, C++, Make, and the S language for graphics and data analysis that was the parent of the R language. He provided a vision for how to carry out the process of data analysis. He wrote a lot of papers, but it was his personal engagement that had its greatest influence. He was a subject matter expert in many fields of science and engineering. This enabled him to inject statistical methods into many fields. He also was heavily engaged in public service. He was a consultant for the U.S. federal government in many ways. He was present as a technical advisor in meetings with the Russians on the nuclear test band treaty. In one he stunned all by saying it would be hard to distinguish a seismic event and an underground nuclear test. In the early days of computing he consulted with the famous team of Burks, Goldstine, and von Neumann, and designed the electric adding circuit they used. He serves as a role model for data scientists, demonstrating how critical engagement in analyses of data is for spawning new ideas about methods and computing with data.