UseR 2012 highlights

The eighth annual R user conference, UseR! 2012, has come and gone — and what an event it was! I’ve been to five useR! conferences so far, and each one improves upon the last. This year’s conference at Vanderbilt was the best so far: an outstanding location (my first visit to Nashville, a great city), excellent facilities (the lecture rooms at Vanderbilt were perfect), great organization, and intereresting and relevant presentations from the speakers. For me, I think the biggest change from previous years was the increased focus on industry applications of R, with at least a 50-50 mix of attendees from government and commercial users of R compared to attendees from academia. At the conference I met R users from John Deere, Caterpillar, Redbox, Google, Kaplan, IBM, Oracle, Nationwide Insurance, Kroger, Nestlé, AT&T, the FDA, the US and Netherlands Census Bureaux, and even the Cleveland Indians baseball team. R is clearly on the rise in the commercial sector. I’m preparing a series of blog posts on some of the key presentations and events at UseR! 2012, but for now here’s a snapshot of some of my personal highlights:

• Uwe Ligges’ tutorial on writing efficient and parallel code in R, and learning new tricks for profiling and optimizing R code.
• The RStudio tutorial, where I learned how to manage github-based R projects with RStudio (and that the control-uparrow command searches history).
• David Kahle’s ggmap packagepresentation, where I learned you can combine a Google or OpenStreetMaps map with a ggplot2 chart.
• Seeing the great progress on fast interactive graphics with R, from Yihui Xie’s cranvas package and Di Cook’s ggobi presentation, to Simon Urbanek’s interactive graphics applications based on FastRWeb.
• Yihui’s knitr package, an incredibly useful tool for creating beautiful reports with R, with great integration with RStudio.
• Hiveplots and the HiveR package, a novel way of visualizing complex social network data.
• The amazing keynote from Kevin Coombes about how forensic bioinformatics uncovered the scandal of the Duke cancer trials (reported on 60 minutes), and the corresponding importance of reproducible research.
• How Tim Hesterberg created the dataframe package to speed up R for 500+ R users at Google, and the talk from his colleague Karl Millar on using Google’s big-data infrastructure with R.
• Ryan Elmore scraping baseball-reference.com with R to settle a bet on whether a “perfect” six-pitch innings had ever occured. (None has: the minimum is 7.)
• Garrett Grolemund demonstrating how to make handling dates and times easier in R with the lubridate package.
• The big-data session, seeing R integrated with RHadoop, Greenplum and Netezza.

But a clear favourite for me was seeing Bill Venables’ closing keynote, R Quo Vadis. Bill has made so many contributions to the R community, from the MASS package and book (with Brian Ripley), to contributing the Introduction to R manual, and providing cheerful advice on the mailing lists. I also learned in the keynote that it was Bill that first convinced Frank Harrell and R-core member Brian Ripley to switch from S+ to R. But because I owe so much personally to Bill — it was he started me on my carreer by convincing me to specialize in Statistics instead of computer science, and who set me off on my adventures with the S (and then R) language — it was especially wonderful to see him presenting at useR! again. The next conference, UseR! 2013, will be held at University of Castilla in La Mancha, Spain. If it’s anything like this years, it will be awesome. See you there!