The goal of in-database analytics is to bring the calculations to the data, reducing transport costs and I/O bottlenecks. With Procedural Languages such as PL/Python and PL/R data parallel queries can be run across terabytes of data using not only pure SQL but also familiar Python and R packages. The Pivotal Data Science team have used this technique to create fraud behaviour models for each individual user in a large corporate network, to understand interception rates at customs checkpoints by accelerating natural language processing of package descriptions and to reduce customer churn by building a sentiment model using customer call centre records.
http://www.meetup.com/Data-Science-Amsterdam/events/178974942/