#37 Prophet, Time Series & Causal Inference, with Sean Taylor

Learning Bayesian Statistics

Episode sponsored by Tidelift: tidelift.com

I don’t know about you, but the notion of time is really intriguing to me: it’s a purely artificial notion; we humans invented it — as an experiment, I asked my cat what time it was one day; needless to say it wasn’t very conclusive… And yet, the notion of time is so central to our lives — our work, leisures and projects depend on it.

So much so that time series predictions represent a big part of the statistics and machine learning world. And to talk about all that, who better than a time master, namely Sean Taylor?

Sean is a co-creator of the Prophet time series package, available in R and Python. He’s a social scientist and statistician specialized in methods for solving causal inference and business decision problems. Sean is particularly interested in building tools for practitioners working on real-world problems, and likes to hang out with people from many fields — computer scientists, economists, political scientists, statisticians, machine learning researchers, business school scholars — although I guess he does that remotely these days…

Currently head of the Rideshare Labs team at Lyft, Sean was a research scientist and manager on Facebook’s Core Data Science Team and did a PhD in information systems at NYU’s Stern School of Business. He did his undergraduate at the University of Pennsylvania, studying economics, finance, and information systems. Last but not least, he grew up in Philadelphia, so, of course, he’s a huge Eagles fan! For my non US listeners, we’re talking about the football team here, not the bird!

We also talked about two of my favorite topics — science communication and epistemology — so I had a lot of fun talking with Sean, and I hope you’ll deem this episode a good investment of your time

Our theme music is « Good Bayesian », by Baba Brinkman (feat MC Lars and Mega Ran). Check out his awesome work at https://bababrinkman.com/ !

Thank you to my Patrons for making this episode possible!

Yusuke Saito, Avi Bryant, Ero Carrera, Brian Huey, Giuliano Cruz, Tim Gasser, James Wade, Tradd Salvo, Adam Bartonicek, William Benton, Alan O'Donnell, Mark Ormsby, Demetri Pananos, James Ahloy, Jon Berezowski, Robin Taylor, Thomas Wiecki, Chad Scherrer, Vincent Arel-Bundock, Nathaniel Neitzke, Zwelithini Tunyiswa, Elea McDonnell Feit, Bertrand Wilden, James Thompson, Stephen Oates, Gian Luca Di Tanna, Jack Wells, Matthew Maldonado, Ian Costley, Ally Salim, Larry Gill, Joshua Duncan, Ian Moran, Paul Oreto, Colin Caprani, George Ho, Colin Carroll, Nathaniel Burbank, Michael Osthege, Rémi Louf, Clive Edelsten, Henri Wallen, Jonathan Sedar, Hugo Botha, Vinh Nguyen and Raul Maldonado.

Visit https://www.patreon.com/learnbayesstats to unlock exclusive Bayesian swag ;)

Links from the show:

• Sean's website: https://seanjtaylor.com/
• Sean on GitHub: https://github.com/seanjtaylor
• Sean on Twitter: https://twitter.com/seanjtaylor
• Prophet docs: https://facebook.github.io/prophet/
• Forecasting at Scale -- How and why we developed Prophet for forecasting at Facebook: https://www.youtube.com/watch?v=OaTAe4W9IfA
• Forecasting at Scale paper: https://www.tandfonline.com/doi/abs/10.1080/00031305.2017.1380080?journalCode=utas20&
• TimeSeers -- Hierarchical version of Prophet, written in PyMC3: https://github.com/MBrouns/timeseers
• The Art of Doing Science and Engineering -- Learning to Learn: https://www.amazon.com/Art-Doing-Science-Engineering-Learning/dp/1732265178
• NeuralProphet -- Forecasting model based on Neural Networks in PyTorch: https://github.com/ourownstory/neural_prophet/
• Introducing PyMC Labs: https://www.pymc-labs.io/blog-posts/saving-the-world/