Predicting road traffic density using a machine learning-driven approach

Notwithstanding the technological developments in transportation systems, traffic congestion is still hampering the growth and development of countries. Accurate traffic flow prediction plays an essential role in intelligent transportation systems to mitigate traffic congestion problems. Importantly, it provides prior knowledge on traffic status, which enables avoiding congested points. This paper employed a Support vector regression (SVR) approach, a kernel-based learning model, to predict traffic flow. We assessed the efficiency of the SVR model for traffic density prediction by considering different types of kernels. We used traffic data from California highways to test the SVR prediction performance. Results showed that SVR with Gaussian kernel dominates the other SVR models.