Bayesian small area estimation with iNZight

A prototype for a Bayesian demographic modelling module using iNZight's add-on system.

Bayesian demography is an important technique for minority communites, and requires special methods to infer the correct results. A set of existing R packages provide a way for researchers to fit hierarchical Bayesian models to estimate birth rates, life expectancy, or forecast important values such as diabetes rates.

iNZight is a tool for exploring and analysing data, and provides a way for researchers low on time, money, and/or skill to quickly generate the necessary results with minimal effort. It also provides the capabilities to do some basic data wrangling, including aggregation of data into an array of counts, necessary for fitting demographic models.

We have developed a prototype for a Bayesian small area estimation module within iNZight, allowing users to create the necessary aggregated array and fit a set of simple models to this data. The results of the MCMC simulation are displayed in the graph automatically, and forecasts and other posterior estimation procedures can be carried out — all using simple point-and-click mechanisms.