6:30 PM - Pizza and Networking
7:00 - Announcements
7:05 - Lightning Talk - Sashikanth Chandrasekaran: snap-analytx
7:20 - Alex Brown: Shiny
8:00 - Justin Hemann: CRM Models in R and Causata / R integration
High quality prediction algorithms are freely available in languages such as R. Yet, predictive analytics remains confined to a few large companies in select industries and modern internet companies. We believe this gap exists because of a shortage of scientists and the complexities involved in deploying the prediction models in a production application.
This talk presents a new approach - nicknamed snap-analytx - to deploying prediction models written in languages such as R. Instead of developing one-off custom prediction models, R programmers can use the new platform to make their prediction models available to a larger number of customers by creating reusable models. The platform relieves the R programmers from system issues surrounding model deployment, data integration, run-time scoring and performance monitoring while still facilitating collaboration with customers in order to tweak or extend their model.
Bio: Sashikanth Chandrasekaran
I am a software engineer currently at Great Bridge Corporation where we use consumer transaction data to predict retail trends. Previously, I was at Google working on dynamic search ads, a new method of automatically creating search ads from a web site's content. I started my software career at Oracle, where I implemented several features in the Oracle database.
Alex started with graphics on "Harvard Presentation Graphics" in the 80s, and has been using R for about 10 years as a magic calculator and to understand data - and share that understanding with colleagues. Alex says:" Most of all I love the graphics it produces".
Abstract for Alex's Talk
Shiny (from the makers of RStudio and the author of plyr, ggplot2 and others) takes your R programs and techniques and throws them into the intra or internet.An R veteran will show you how to get started, and explain how shiny enables you and your team to share discoveries, activate analyses for interaction, link your graphs to live data and more.
Abstract: Justin's Talk
Suppose that you are a data scientist, and you've been asked by your boss to make the company website more personalized and relevant to visitors. You decide that the best course of action is to try to predict what visitors will do, then show content related to the prediction on the website. Causata was created to solve this kind of problem, which involves three major steps: collecting data into a single view of the customer, learning from the data, and taking action. A brief demo will show how Causata and R are used together. Then the focus will move to customer segmentation, creating interesting features in the data, and useful R packages.
Bio: Justin Hemann
Justin Hemann is a Principal Data Scientist at Causata and the primary author of the Causata R package. Justin uses R to create predictive models that drive decisions for customer experience management, including web content, offers, and emails. Justin's background includes predictive modeling around customer experience management, marketing mix models, aviation noise modeling, and orbit dynamics. Justin has an MS in Engineering Systems from MIT and a BS in Aerospace Engineering from Purdue.