Most data engineering challenges are concerned with processing large amounts of raw data and converting it to structures that are convenient for counting, machine learning, and visualization. At Generable our challenges are at the other end of the pipeline.
We are building high dimensional, generative models, and post-processing model inferences for making decisions under uncertainty. We have not solved this problem in full generality, but we have learned about some of the unique challenges. We will demonstrate our process and highlight these challenges using inferences from Bayesian Survival models that are often used to assess efficacy and safety of new therapies during and after clinical trials.
Eric Novik is an applied statistician and a founder and CEO of Generable (formerly Stan Goup). Prior to Generable, Eric was a Data Scientist at TIBCO Spotfire where he built statistical applications for customers in Financial Services, Energy, Pharma, and Consumer Goods Sectors. In 2010, Eric started Risktail, an options analytics software company for retail traders. During the financial crisis of 2007-2009, he traded listed options and later spent two years in the Equity Derivatives Technology group at Barclays Capital. He has an MA in Statistics from Columbia University.