A. Hanea and W. Harrington

Ordinal Data Mining for Fine Particles with Non Parametric Continuous Bayesian Belief Nets


We introduce a Bayesian Belief Net (BBN) based approach for analysing the relationship between SO2 emissions, and concentrations of fine particulate matter, PM2.5. PM2.5 exposure has adverse health effects, hence we study this relationship with the goal of quantifying the health benefits of emission reductions. The main advantage of our approach is that it can handle a large number of continuous variables, without making any assumptions about their marginal distributions, in a fast manner. Rapid computations are of little value if the model itself cannot be validated. We discuss the issue of validation and additionally perform backward and forward inference.

КЛЮЧЕВЫЕ СЛОВА: Bayesian Belief Net, validation, health effects