Okay

so the example that we're going to work through is actually this has come up a few times a few people have asked me about.

Do you would you always use a Gaussian distribution as your prior or other scenarios in which you use a different distribution and the answer is yes and one other thing that I just want to.

I might have this wrong, but let's all in the same room try to be wrong in the same way.

So what I'm calling a prior is the distribution that we are inferring.

hierarchically.

And what I'm calling a hyper-prior is the prior one level above that, where we're not inferring anything.

I want you to come out of this session with an understanding of what a hierarchical model is and why it might be useful for your research going forward.

Hierarchical Bayesian modeling is really quite jargony, and it in and of itself is not very complicated.

There's really just one main concept that if you get that, you'll understand what it is that we're doing.

And it's one of these things that you have to be kind of pragmatic over it.

Like sometimes it's really useful to create a hierarchical Bayesian model.

But sometimes it's too much computation time for what it's worth.

We'll see.

OK, so in PyMC3, fitting a line to data.