Earlier this month I participated in rstudio::conf 2018 in San Diego (many, many thanks to RStudio for that opportunity!). I learned practical skills, got ideas from so many people in different contexts, and (best of all) was able to both reconnect with and meet so many R-world folks. I wanted to sum up some of my favorite takeaways for my own notes, but also for anyone thinking about attending. (TL;DR: You should.) In bulleted list form and no particular order:
Empathy was all over this conference. Most obvious was JD Long’s great talk on “The Unreasonable Effectiveness of Empathy” (previous version), in which he talked about how putting ourselves in someone else’s shoes can be helpful in both communicating results and in anticipating the needs of our clients/collaborators/users (thus saving ourselves time and pain). Kayla Patel talked about developing a Shiny dashboard for Imagine Boston 2030 that keeps the end user in mind first and foremost - for example, prioritizing clarity and accessibility over technical bells and whistles. At the closing fireside chats, Eduardo Ariño de la Rubia had some important thoughts on viewing hiring from an employee’s perspective and workers’ identity in a changing society.
I’ll stop here, but I could go on. It was so evident that the group gathered cares about understanding others’ experiences, from well designed tools to creating community to making our work have an impact on our consumers.
This conference brought people together from all kinds of backgrounds and contexts - just off the top of my head, I met people with backgrounds in marine science and ecology, GIS, energy, tech, and agriculture. Everyone was learning and being inspired by each other, while having a great time doing it. Only good can come from this.
The conference very intentionally had a welcoming, community-focused vibe, starting beforehand with the “Birds of a Feather” group announcements and going all the way through the weekend. The program was built so that everyone had somewhere to go and something to learn, and having everything in one area created plenty of chances for spontaneous conversation. The friendly, approachable environment, focus on connections, and genuinely fun people made the whole experience better.
People were encouraged over and over to take what they’d learned and immediately contribute to the community or to their workplace. Several of the RStudio team’s talks focused on how to contribute to R, whether via writing your own packages; contributing to other people’s, as Mara Averick encouraged us to do…
Great point from @dataandme on reasons to bite the bullet and contribute to packages: Users submitting fixes for "small" things like typos frees maintainers to focus on "big" things like new features. #rstudioconf— Jennifer Thompson (@jent103) February 3, 2018
…or taking new skills and and ideas and improving your own workflows and practices. As much or more than “look at the cool stuff [Well Known Person] does,” people seemed to feel really excited about what they could do and how they could collaborate and contribute.
Every talk I went to was about something powerful in its application - whether that was presenting analysis to the end user in a way that was clear and communicative, or a tool to make [data people’s] lives easier, or a way to teach and empower other people. The sessions weren’t overly technical in an attempt to impress; they were full of excitement and ideas for future work. I can’t really say it better than JD:
After sitting through JJ’s Deep Learning prez at #rstudioconf I realize why this conf feels different than most tech confs: No bullshit. No hype. Just solutions to problems with honesty around the limits of our tools. This is so very refreshing.— JD Long (@CMastication) February 3, 2018
That focus on useful, impactful tools and case studies helped me feel like I could immediately contribute to the community and improve my own work.
With machine learning and prediction so hot right now, it was great to hear plenty of conversation about statistical inference as well. Di Cook gave an engaging and practical keynote that focused on the role visualization can play in inference. Andrew Bray gave a talk on the
infer package, which intends to make statistical inference more transparent whether using approximation or computational methods; I’ll be interested to keep an eye on this package as it grows.
- Di Cook’s keynote on the role of visualisation in inference
- JD Long’s “The unreasonable effectiveness of empathy”
- Ways to improve workflows and collaboration:
- Sandra Griffith of Flatiron Health and Emily Riederer of CapitalOne both talked about the process of growing an R-centric environment at their workplaces (Emily’s slides). Both these women have worked hard to get their teams using a consistent R framework, and they have great thoughts and ideas about how to start and maintain that environment.
- Elaine McVey talked about incorporating Agile software development practices into her data science workflow (summarized here and here).
usethispackage has been around awhile, but between Hadley’s workshop and Jim Hester’s talk on creating an R package in 20 minutes, I’m convinced that I need to dig more into it and keep an eye on what the RStudio team is adding.
- Carson Sievert blew everyone’s minds with his talk on interactive exploratory data analysis. I’m excited to dig into his code.
- Work from Edzer Pebesma on spatial data, Thomas Lin Pedersen on network analysis, and Davis Vaughn and Earo Wang on time series tools made me really excited to learn more about those areas. (Earo wrote a blog post comparing the
- Thanks to Hadley’s “Extending the Tidyverse” workshop (which I would wholeheartedly recommend), I have now officially written approximately five lines of working tidyeval. It was very exciting.
- I got to wear my Chacos in the dead of winter, and I’ll be forever grateful. Thanks, San Diego.
Again, many MANY thanks to RStudio for the opportunity to attend this year! The entire experience was both fun and incredibly useful. Next year is already in the works…