Credibility of various sources of data for use in models
The development of reliable models requires eliciting, calibrating, and adjusting for bias in event probabilities that could impact decisions for early warning systems (EWS). This shall be addressed in this project via probabilistic models of uncertainty, specifically, Bayesian networks with probabilistic inference mechanisms. Data credibility can be included in these models and assessed by calculating risk and biases using various measures of uncertainty.
- Co-Project Investigators: Bouchra Nasri (Université de Montréal), Hélène Carabin (Université de Montréal), Julien Arino (University of Manitoba) and Jane Parmley (University of Guelph)