Use of machine learning techniques in road safety modeling: mapping study

Abstract

The road safety modeling is an important alternative in the optimization of resources and efforts to promote safe mobility. This paper presents the mapping study of papers about the development of accident prediction models, especially on highways, using machine learning (ML) techniques. For this purpose, a revision management protocol was applied, using the Portal of Periodicals Capes and Google Scholar as databases. Initially some bibliometric aspects were presented, followed by a qualitative analysis. As a result, the main methodological approaches and their characteristics, model performance and explanatory variables were identified. In this way, the mapping was important to draw the panorama of the area of research, to point out limitations and opportunities of investigation and also, to highlight the potential of the use of ML for analysis of crash accidents.