Title: Reti bayesiane per lo studio del fenomeno degli incidenti stradali tra i giovani in Toscana

Abstract: This paper aims to analyse adolescents' road accidents in Tuscany. The analysis is based on the Database Edit of Osservatorio di Epidemiologia della Toscana. Complexity and heterogeneity of Edit's data represet an interesting scope to apply Machine Learning methods. In particular, in this paper is proposed an analysis based on a Bayesian probabilistic network, used to discover relationships between adolescents' characteristics and behaviours that are more often associated with an audacious driving style. The probabilistic network developed by this study can be considered a useful starting point for follow up reasearches, aiming to develop a causal network, a tool to limit this phenomenon.