Linear regression with Generalized extreme value distribution

Can you just check a couple of things?

I think the exponent here should be **, otherwise you’ll get a bitwise operation.

μ = alpha + beta*X +gam*X^2
ξ = pm.Normal("ξ", mu=0.0, sigma=0.35)

This is a wide prior for ξ and could be giving trouble - trying reducing it to maybe sigma=0.1 and if that runs, bring it back up gradually then to a suitable prior.

2 Likes