Control Theory and Bayesian Networks for Black Spots Study

Traffic accidents are recurring ordinary disruptive events in an urban road network. To measure the performance and recovery of an urban road network, it is necessary to identify and predict the locations on the network where they occur. The application of control theory to identify black spots has mainly focused on interstate and inter-regional highways. However, there is limited literature on the application of these theories to urban road networks and in the definition of black spots. This article introduces an innovative methodology for defining black spots based on control theory, which is adapted to urban road networks with appropriate measurement metrics that prove more suitable than those used following Euclidean norms. To predict whether a traffic accident occurs at a black spot, a Bayesian network is employed, which also indicates the reliability of the accident classification result. It demonstrates better efficiency than methodologies based on logistic regression and Naive Bayes, which is why this algorithm is chosen. The article concludes by measuring the impact of the most significant black spots in an urban road network in Metropolitan Lima, using the concept of transport resilience.