Conceptual Gaussian Process Questions

[willtebbutt]

Just so that Turing gets some air-time in this thread, here’s a Turing + Stheno example that I wrote other day here:

Gaussian Process Model with Turing

Firstly, many apologies for taking so long to reply. So it looks like the following works just fine: Turing.jl + Stheno.jl implementation using DataFrames using Distributions using Random using Stheno using Turing using Zygote # Use Zygote for AD. Be sure to use 0.4.20 rather than 0.5. Turing.setadbackend(:zygote) # Generate fake data. df = DataFrame( subject = repeat(1:5, inner=3), obese = repeat(rand(Bool, 5), inner=3), timepoint = [1,2,3,1,3,4,1,2,5,1,4,5,1,3,5], bug = ran…