by Joseph Rickert
Earlier this month the Bay Area useR Group (BARUG) held it annual lightning talk meeting. This is by far our most popular meeting format: eight, 15 minute talks (12 minutes speaking and 3 minutes Q & A while the next speaker is setting up) packed into a two hour time slot. The intensity seems to really energize the speakers and engaged the audience.
Bradley Shanrock-Solberg kicked off the event with delightful example of an R Monte Carlo simulation based on his wildpoker package that you can find on CRAN. I have never seen a more prepared lightning talk presenter: high energy, a royal flush presentation and a four color printed hand out just in case you have trouble keeping up with him for the 12 minutes. In a series of well conceived plots Bradley showed how, for a number of different poker variations, the best hand changes as the game progresses. The number of players who start the game, the number who stay until the showdown, wildcards and many more contingent events dynamically change the value of your hand. Bradley is definitely the guy for your next trip to Vegas.
William Sundstrom, professor of Economics at Santa Clara University, gave an entertaining and thought provoking presentation on teaching R to undergraduate Econometrics students. One interesting observation that generated some discussion was that even though today's students are "digital natives" having grown up using intelligent devices of all kinds, many of them are nevertheless "digital naïfs". The following slide, a reprint of an email from one of Professor Sundstrom's students, captures some of their frustration.
David Ouyang MD, a Stanford resident, a guy who sometime puts in 73+ hour work weeks, presented some explorations of the Epic electronic medical records data set he is analyzing in his spare time. The following plot shows the distribution of physician's interactions with the Epic system over the course of a day.
Dennis Noren, a long time BARUG member and contributor showed some results from a recommendation App he is building on top of The Movie database. The following slide from his presentation shows a Shiny dashboard he build to drive a parallel plot. The interactivity really makes the plot useful.
If you are thinking about betting on the Academy Awards you might want to consider Keith Everett's predictions based on his GLMNet Model.
Nelson Auner talked about Modern KPI Tracking in R. My favorite slide describes the behavior of Execs who don't work for companies that sell BI Software:
Many thanks to Earl Hubbell and the folks at Thermo Fisher Scientific who hosted the meeting.