Querying bayesian model output with PostgreSQL

A presentation to the Wellington PostgreSQL Users Group, about the use of PostgreSQL for storing Bayesian model output. Samples from the posterior distributions are stored in arrays, allowing for rapid aggregation of the results. In our case, models of protected species bycatch have 4000 samples for each of 1.5 million fishing events. These may be aggregated by area, or by fishing method, to give estimates of bycatch with the associated uncertainty. This database of model output is used to generate the graphs and tables seen on the Protected Species Bycatch website.