Citation

Arnaldo Valdés RM, Gómez Comendador VF, Perez Sanz L, Rodriguez Sanz A. Safety Sci. 2018; 104: 216-230.

Copyright

(Copyright © 2018, Elsevier Publishing)

DOI

10.1016/j.ssci.2018.01.008

PMID

unavailable

Abstract

Today, aviation is immersed in a shift from old-fashioned reactive and compliance-based safety approaches towards proactive and performance-based methods and tools. Stakeholders have to monitor, gather and analyse safety-related data and information in order to anticipate and predict actual and emerging safety risks. In this context safety analytics and statistics need to evolve to forecast future safety performances and risks. This research adopts an innovative statistical approach involving the use of Bayesian inference and Hierarchical structures to develop statistical estimation and prediction models with different complexities and objectives. The study develops and analyses five Bayesian models of increasing difficulty, two basic and three Hierarchical models, which allows us to explore safety incident data, efficiently identify anomalies, assess the level of risk, define an objective framework for comparing air carriers, and finally predict and anticipate incidents.

Language: en