A transformative deep learning framework for traffic modelling using sensors-based multi-resolution traffic data

Intelligent transportation system (ITS) is a cohesive organisation of roads, vehicles, and people which utilised the power of computing for the management of unpredictable traffic conditions. Today, road accidents and congestion are the major problems that arise due to absence of ITS. Due to intricate and dynamic spatio-temporal linkages between various regions in road network, particularly at major intersections of city, these issues are difficult to solve. Deep learning offers enormous potential to enhance traffic operation and management when combined with current sensors-based multi-resolution traffic data and future linked technologies. But we require effective modelling frameworks for deep-learning algorithms to address complicated transportation problems. This work mainly focused on the analysis of traffic conditions using deep learning hybrid framework for a correct prediction of traffic pattern with a real world traffic dataset. Performance of proposed framework betters the benchmarks with RMSE of 52 and MAE of 49.