Data-driven traffic incident detection in urban roads based on machine learning algorithms

Abstract

Known issues such as traffic congestion, pollution, and travel delays are mainly caused by incidents in urban roads, so incidents need to be detected for better management. This paper describes various machine learning algorithms for incident detection, like Support Vector Machine (SVM), Random Forest (RF) and long short-term memory network (LSTM). To assess the effectiveness of these models, simulated data were generated through the utilization of the open-source software SUMO. And the obtained results show that the LSTM achieve a good performance when it’s compared to SVM and Random Forest.