More than
100,000,000
Models built by datarobot customers

DataRobot at R User Conference 2014: Highlights and Presentation Slides

July 8, 2014

We just wrapped up a busy week in Los Angeles at the R User Conference 2014. The four-day event was very well done, and the quality of attendees we met throughout the conference was outstanding. This is an event that will definitely be on our annual calendar!

DataRobot and R User Conference

Our own Xavier Conort spoke about Ten R Packages to Help Win Kaggle Competitions. In his session, Xavier talked about his journey from an actuary to a machine learning expert and how Kaggle competitions and R played a key role along the way. Xavier shared the top R packages he used to win Kaggle competitions, and tips for data scientists to build better predictive models.

Highlights of the Session

These are some lessons from competitions Xavier participated in:

  • Machines can do a better job at capturing complexity in data than humans
  • Don’t oversimplify and discard any data prematurely
  • Feature selection can play a role in getting to the right model
  • Parallel processing and big servers can make a huge difference
  • Take advantage of high-cardinality categorical or text features
  • Use your intuition to help the machine do its job better
  • There are so many useful R tools out there and so much more to learn!
  • You learn something new every competition!

 

View the entire presentation below:

In addition to presenting, we also exhibited at the conference. We gave out DataRobot t-shirts and showed some attendees a sneak peek of the DataRobot platform. If you’re interested in learning more about DataRobot, request an invite to our private beta.

If you attended the conference, what did you find most interesting or insightful about the growing R community and conference this year?