Identification of rules induced through decision tree algorithm for detection of traffic accidents with victims: a study case from Brazil

Abstract As traffic demand continues to increase and in the face of limited possibilities to expand the road infrastructure, various attempts are being made to make road traffic safer. In order to decrease the number of traffic accidents, it is required knowledge of its causes, so that actions are taken to avoid them. This is possible through collection and management of the available information on traffic accidents. Thus, policies are needed to identify and manage all available information relating to traffic accidents. This study aims to identify rules induced by Decision Tree algorithms (DT) for detecting traffic accidents with victims in a road stretch from accidents records, as well as probable causes of the occurrence and type of accident. Data are from a road stretch of the Regis Bittencourt (highway BR-116) between km 509 to km 519 in the period 2012–2014, located in Sao Paulo, Brazil. Through the main results obtained, it can be concluded that the CART algorithm of the Decision Tree is a useful tool in identifying potential sites of accidents with victims. In this case, the two most important variables to identify the severity of accidents were the accident type and the accident cause.