missed meeting / Stan short course

[Bob Carpenter]

I was teaching a 4-hour (R)Stan course at UTS
Sydney yesterday (organized by Louise Ryan through
the Stat Society of Australia, NSW) and then went out with
people afteward and completely forgot the late-night
meeting.

The course went OK, but I had way too much material (typical
rookie mistake, as Mitzi pointed out). People seemed to
enjoy the practical demos more than the more lecture-oriented
material --- at least it invoked more question. I'd like to
make it more hands-on.

I have a long way to go in figuring out how to present
Stan to a mixed audience in 4 hours. The crowd both wanted
to know how to program in Stan and also how it worked under
the hood, but some of them only had a thin background in
Bayesian stats.

Daniel, Andrew, and I are going to be doing a three-day
course (the plan is to make it modular, with me doing a one-day
intro to Bayes and MCMC, Daniel doing a one-day intro to Stan,
and Andrew will do a one-day session on Bayesian applications
drawn from his deep well of talks --- attendees will be able to
choose which subset of 3 days to attend).

- Bob

[Daniel Lee]

On Tue, Mar 24, 2015 at 7:21 PM, Bob Carpenter <ca...@alias-i.com> wrote:

I was teaching a 4-hour (R)Stan course at UTS
Sydney yesterday (organized by Louise Ryan through
the Stat Society of Australia, NSW) and then went out with
people afteward and completely forgot the late-night
meeting.

No worries. We started off slow, but ended up talking about the callback and making progress. I think we're all on the same page: we'll implement with callbacks now. The writers will change to functors (instead of 3 function signatures, which might expand to more later, it'll just be operator()).

 

The course went OK, but I had way too much material (typical
rookie mistake, as Mitzi pointed out).  People seemed to
enjoy the practical demos more than the more lecture-oriented
material --- at least it invoked more question.  I'd like to
make it more hands-on.

Yup. I think as we do more of these, we'll get a better sense of how to balance lecture and hands-on.

 

I have a long way to go in figuring out how to present
Stan to a mixed audience in 4 hours.  The crowd both wanted
to know how to program in Stan and also how it worked under
the hood, but some of them only had a thin background in
Bayesian stats.

Daniel, Andrew, and I are going to be doing a three-day
course (the plan is to make it modular, with me doing a one-day
intro to Bayes and MCMC, Daniel doing a one-day intro to Stan,
and Andrew will do a one-day session on Bayesian applications
drawn from his deep well of talks --- attendees will be able to
choose which subset of 3 days to attend).

I'm looking forward to it. Is it wrong that I find this stuff fun?

 

[Bob Carpenter]

I like teaching, too! I hate the prepping for it, though.
I prefer writing in long form to writing in slide form.

And I'm still bummed that I feel like I really messed this one up
in terms of level and hands-on-edness (made it too basic then rushed
through the advanced stuff and didn't do nearly as much hands on as
advertised --- I did demos but didn't have exercises for students; I
should've done 1/5th as much material and had it mostly be hands on
with Stan).

- Bob

> --
> You received this message because you are subscribed to the Google Groups "stan development mailing list" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to stan-dev+u...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.