Employing Bayesian Networks and conditional probability functions for determining dependences in road traffic accidents data

As we can all experience in our daily life, the traffic in the cities grows quickly, which, unfortunately, means also that the number of accidents grows, too. We try to find causes of accidents that happen for systematic reasons as we perceive eliminating such systematic errors as one of primary goals of smart cities idea. This paper deals with the accident data analysis using Bayesian Networks and conditional probability functions. We try to examine independence between variables in data sample in order to work with data of considerably large dimension. Our approach includes determining the structure of a Bayesian Network basing on a data sample and then utilizing computed probabilities in order to eliminate insignificant relations. We also use conditional probability functions to identify significant dependences basing only on data set. Finally we compare results obtained by both methods and use Goodman and Kruskal's lambda coefficient for confirming their accuracy.