Oscar Padilla | Hadoop Summit 2014
summary
Uses Hadoop as a central data lake and has used it since 2012 to build empirical audience models for the Latino market, ingesting about 2,000 sources and processing roughly 15 TB/month. Latency drops from three days to about three hours, workflows run R and Tableau and produce SaaS analytics, and a hub‑and‑spoke architecture surfaces correlations from transactional POS, loyalty and online/offline consumption data to deliver targeted analytics at scale.
AI-written summary · about
