John Salvatier: Bayesian inference with PyMC 3

[@WalterReade]

Great video. I've been impressed with the number of variables PyMC3 can handle (I've created models with 2,000+). Theano compilation time can get very long, though, requiring more time than the actual sampling at times.

[@stepkurniawan]

Is there any github link to point out to the presentation?

[@michaelmoore7568]

Allen Downey recommends a half-Cauchy for Nu, and sigma... maybe, i am misunderstanding something though.

[@ChrisMNetrunner]

Thanks for the video. I had to stop listening and look at the slides though because of how often you said 'Uh' and 'Um'. Something to work on :)

[@qrubmeeaz]

The 'aaa's are very distracting.