I’d attend! And try to help otherwise also … @hneyster@maxfarrell@legault@Deirdre_L@AileneKane@FaithJones might be interested or know folks who would be also. As a start I’d like to learn more about the multi-species occupancy models @hneyster mentioned recently.
Hi all,
Since it seems like there’s interest in doing this, I have a couple of questions for you and @SGB. A proposal is due on 26 April.
I need to select a platform and provide a cost estimate to @SGB. However, I will be changing institutions on 1 May, and so I don’t really know the details of what sort of Zoom account/policies I will have. Is anybody willing to co-organize by taking the lead on providing/costing the platform? Alternatively can @SGB wait until mid-May for details of the platform and a cost estimate?
Organizers of the Cognitive Science StanConnect put out a public call for abstracts to be voted on by the community for acceptance. I could do the same, or I could hand-pick speakers. I think both options have advantages and disadvantages. Does anyone (including @SGB) have strong preferences here? Advantages of a public call for proposals and public voting include transparency, publicity, and ability to reach a wide variety of speakers, with resulting benefits for diversity and inclusion. Disadvantages include potential problems for diversity and inclusion depending on the vicissitudes of the voting, wasted abstract-writing effort of people whose abstracts aren’t accepted, the difficulty of translating voting into a coherent and well-balanced set of talks, and the possibility that we won’t attract enough submissions in the short available timeframe.
If a community consensus in support of a public call doesn’t emerge by Wednesday evening of this week, I will hand-pick the speakers with an explicit commitment to diversity and inclusion and with a focus on using Stan for ecological models that are difficult to fit (or difficult to verify the validity of the fit) using alternative tools.
We need to pick a date. July or September/October work best for me. I know that July might run into field seasons for some, while September/October might run into teaching/coursework obligations. If you think you might attend (even if not sure), please fill out the poll below (at the bottom of this post). If I end up hand-picking the speakers, I will balance the poll results against the speakers’ needs.
We need to pick a time. As far as I know, the majority of people who have expressed interest are in the Americas, and so I suggest noon EDT as a time that would accommodate most people, including interested people in Africa and Europe.
Unless there are strong objections, I will not follow the StanCon tradition of requiring that data and analyses be made available in permissively licensed markdown or Jupyter notebooks. This is to allow speakers flexibility in presenting ongoing work and to encourage participation from ecologists who don’t necessarily write the cleanest of code.
It’s fine to wait till mid-May to make decisions on platform, please indicate that in the submission.
SGB is agnostic on how organizers coordinate but would like to focus on the submitted proposal. Keep in mind that within logistic limit there’s no hard cap on how many StanConnect events we can have in a year, and if we have multiple proposals in the same area we’d be thrilled and would do our best to accommodate everyone. Personally I’d avoid voting on abstracts.
Just to add, I think it might be sensible to also consider conferencing platforms that allow for networking (this is basically allowing sessions where participants meet at virtual tables) - here’s what I wrote on the topic when discussing this topic with SGB.
Not sure how much you discussed this, but I have not found a free solution for online events that would work good for networking. I liked Remo.Co that hosted the past StanCon, but it seems pricey (USD 450/850 depending on having < 200 people). AirMeet is cheaper (USD 100/250 depending on having < 100 people), while still reasonable feature-wise in my experience.
Additionally, I am partially affiliated with the Elixir infrastructure for bioinformatics, where an ecology StanConnect might be almost in scope and they do have licenses for larger gatherings via GoToMeeting (which is not great, but also not terrible) So if necessary, I could inquire whether they’ll be open to provide the platform in exchange for showing their logo and StanConnect being listed on the Elixir web. (I guess they would gladly accept as any activity that can be claimed as “output” is usually very welcome).
I’d be interested in attending. I’m currently collaborating on an ecology-related project and expect to be involved in more such projects in the future. Nothing in a state that I could submit anytime soon though :-/
Hi all, September/October is a clear winner, and nobody seems to mind if I hand-pick speakers.
Nevertheless, please reach out if you are interested in presenting (and haven’t already commented to that effect on this thread)! I am especially interested in talks from anybody who has done work that would have been much harder to do with commoner tools for ecological data analysis. This can include (but is not limited to):
models that would have been difficult to code in other software but that are readily fit via brms or rstanarm
models with gobs of parameters that are difficult to fit with JAGS (i.e. models that take real advantage of HMC’s scalability to high dimensions)
models where the availability of divergences as a diagnostic caught important problems that would have been missed otherwise
analysis pipelines that take advantage of other parts of the Stan ecosystem in R (e.g. loo or projpred)
exotic models like ODE models and so forth
I will pick a date in the fall based on the schedules of the speakers. @SGB is explicitly open to multiple StanConnects around similar topics, so if there is an excess of speakers I will consider organizing a two-parter.
This all sounds fantastic! I will definitely attend but can’t imagine I’d have anything ready worth presenting. I can’t really help with the platform stuff given my current position but can help coordinate with emails and other logistical stuff as needed.
I’d definitely attend. I may have some work to present that others might find interesting. I’ve been translating the JAGS models for the North American Breeding Bird survey into Stan (while simultaneously translating my workflow to Stan). Large improvements in sampling efficiency for the base models, able to fit more complex models, and some neat spatial and temporal elaborations that weren’t possible in JAGS (e.g., incorporating some of the cool work that @mitzimorris and others did with spatial CAR structures to better share population-change information across a species’ range).