The Paradox of Big Data and Predictive Modeling
Tales from the Data Trenches of Display Advertising
As Chief Scientist at Dstillery, Claudia Perlich works to collect about 10 billion user events daily, representing the digital and geo-physical journeys of millions of people on desktops, tablets, and mobile phones. In this lecture, Perlich explores a number of challenges including privacy-preserving representations, robust high-dimensional modeling, large-scale automated learning systems, transfer learning, and fraud detection. Perlich also touches on a few higher-level lessons and poses the paradox of big data and predictive modeling: You never have the data you need.