Deadline for CFP: September 4th, 2020. Anywhere on Earth.
Date of speaker content submission: TBD. Likely first week of October 2020.
Date of Async Conference Start: TBD. Likely October 23rd.
Date of Synchronous Conference: October 31st.
PyMCon 2020 is an asynchronous-first conference. All accepted speakers will submit a video with their presentation ahead of time.
We are looking for folks like you to share your cool tricks, best practices, general knowledge, or model build methods with the Bayesian community. Presentations should have some relevance to the PyMC community, more details on what we look for in a presentation and how to propose one can be found below.
All presentations will be pre-recorded and will be shared on the PyMC discourse in late October for the async component. We are also planning a synchronous component on October 31st . Further details are below as well.
This conference will have two tracks:
There are three presentation formats. The topics do not have to be about PyMC specifically but something the PyMC relevant. E.g. a new bayesian diagnostic would be acceptable :
PyMCon 2020 is an asynchronous-first conference, that will include a both an async discussion and synchronous discussions. All presentations have three primary components:
Put on a shirt. Pants optional. Hit record You as a speaker will be giving a presentation as you would at any other conference, except that you will be recording it ahead of time. PyMCon will collect the videos and post them on online about a week prior to the conference day.
Put on a shirt. Pants optional. Answer questions on discourse All presentations will be posted onto the PyMC discourse so the community as a whole can engage, read, and comment, all at their own pace. As a speaker, this allows your presentation to be seen by a much wider audience than if just presented once and have a richer conversation with your viewers.
Dress your best and meet all the folks This day will look more like the open source conferences you've loved in the past. As an attendee this day will likely consist of Lightning Talks, Birds of a Feather discussion sessions, and discussion sessions from speakers. We'd like to have speaker available answer questions live for folks in sessions as well. The exact timing is TBD, but we plan to find time zones that are convenient for everyone and may end up doing two synchronous sessions in the same calendar day to meet our global community's localities.
Talks will consist of a 30 to 40 minutes video. As pre-recorded videos, talks are expected to be polished and crisp.
All the content shown in the video (slides, code, plots...) and language should be professional, comprehensible and follow PyMCon code of conduct. A successful talk does not have to be about a novel, cutting-edge technology topic but should be distinctive and have the ability to keep the attention of the audience. We will provide a platform for audience discussions and Q&A. Accepted speakers are asked to be online the day of for at least one session of Q&A to answer questions live, which is 20 minutes long; of course, they are free to participate in the discussions with the audience outside of this Q&A session as well.
Tutorials are dedicated to introducing, promoting, and advancing specific techniques/practices in computational Bayesian statistics with PyMC. They should cover a topic more in depth than a traditional talk. Tutorials will consist of a 60 minutes video and extra material for attendees to engage and experiment with the contents of the tutorial.
Tutorial facilitators are responsible for preparing interactive and carefully designed materials, including hands-on exercises that help reinforce the concepts discussed in their topics. These will be postedto Discourse where tutorial facilitators can communicate and share materials with their audience.
We are excited to introduce a new conference presentation format called "Let’s Build a Model" (LBAM). While sharing some characteristics with tutorials, LBAM talks aim to showcase the exciting (and sometimes messy!) nature of Bayesian modeling.
In a LBAM talk, the presenter should first establish a specific modeling goal and then proceed to build that Bayesian model from scratch using PyMC (e.g., in a Jupyter notebook). Like tutorial, LBAM videos should be 60 minutes long.
We know Bayesian modeling is a considerably involved process and can therefore be daunting to new users. With LBAM talks, we aim to address this problem by encouraging speakers to be as thorough as possible, while making sure to include relevant “tricks of the trade” that they find useful in their own experience.
So many are people are doing so many cools things we don't want to restrict ideas with a "strong prior". However we will consider proposals based on the following principles
Originality Proposals should that originality in some dimension whether that be in presentation approach, visuals, math, whatever! Be creative and show it off! Proposals that are already publicly available or similar to published videos have a low probability of being accepted.
Relevance to PyMC community This doesn't mean you do need to give a talk on PyMC3/4. But it does mean think about the diverse and expansive PyMC community and how your presentation would elevate them.
We're not looking for typos! We recognize that folks come from a variety of backgrounds and English isn't always the first language. We will not be judging the proposals for typos and for accepted speakers who would like language assistance we'll have folks who can help.
Submitting a proposal to PyMCon will require the follwing information about the speaker:
The proposal itself should include the following information:
The review process at PyMCon will be double blind: speakers will not know who are the reviewers and reviewers will not know who has written the proposals they review.
We want to make sure the selection process is as unbiased as possible so that everyone has the same chances of being selected as a speaker at PyMCon. This does not mean however that you won't get any feedback on your proposal, quite the opposite: every proposal will receive feedback from the reviewers (via the anonymizing committee to preserve the double blind), especially if rejected.
We aim to make the submission process both enjoyable and useful for everybody. Thus, even if your proposal was not accepted, you'd still get valuable feedback to help you with your next submission. We would encourage you to submit the extended proposal to another conference or meetup.
Here are some tips and tricks to maximize your chances of your proposal being accepted.
Here are some tricks and best practices to get your proposal accepted.
If that were the case, don't worry, you'll still get feedback on your proposal. We value your time and are honored you considered PyMCon as a place to speak. We will do our best to explain to each author the reasons behind our decision and strive for constructive criticism. We want submitting to PyMCon to be an enjoyable and useful experience for everybody, even if your proposal is not selected, you will receive feedback on your proposal that you can use for next year or to send the extended proposal to another conference.
The tips above were adapted from the PyData and JupyterCon resources. We also recommend checking out blog posts by Craig Kerstiens and Hynek Schlawack for more great advice on this topic.