Interactive Graphics at iHeartRadio using Hadoop, R, and Shiny

In this talk, we’ll discuss how iHeartRadio uses interactive graphics to make big data both available and (more importantly) approachable across all levels of the company, as well as encourage experimentation and rapid prototyping of new features.

In particular, we’ll focus on our recent experiences with using Shiny (http://shiny.rstudio.com/gallery/) - an open source web framework based on the R (http://www.r-project.org/) statistical language. We’ll cover the technical challenges we faced as we adopted the framework - integrating with our data backends (Hadoop, Impala, SQL Server), data processing, flexible and robust deployment, etc. We’ll also present some real-world problems we’ve tackled using Shiny, from authoring simple BI dashboards to building full-fledged, domain-specific analytics platforms. Finally, we’ll discuss how we’ve adopted the framework to build low-cost high-fidelity prototypes, allowing for faster, more pervasive experimentation and leading to better, more robust systems and product features.

Free pizza and beer will be provided, so come hungry and be ready to learn!

About the speaker:

Misha works as a data scientist at iHeartRadio. During his time at the company, he’s worked on a variety of projects, including search and recommendation engines, custom radio algorithms, data visualization platforms and terrestrial radio analysis tools.