Urban Traffic Flow Predictions with Impacts of Weather and Holidays

Abstract

Urban road traffic is influenced by many exogenous factors. Many research works have been carried out to analyze the impact of exogenous factors over traffic flow parameters. Among many other factors, holidays and weather influence urban road traffic to a major extent. Because of such exogenous factors, the traffic flow will be varied from usual traffic conditions due to drivers’ decisions in choosing the mode of transport and time of travel. Hence a significant switch will be found in urban scenarios due to switching from public to private transportation modes during rainy or snowy days. This research work aimed to analyze the effects of weather (rainfall) and holidays on the forecasting of traffic volume within an urban area. One of the state-of-the-art time series prediction models is the Neural Prophet (NP) model (2020). Being quite new in the area of traffic engineering and having more benefits with decomposable additive model, NP model was chosen for forecasting the urban traffic with effect of exogenous variables. Traffic data from a busy urban area in Duisburg city, Germany was used for training and testing the model. The results from this research work showed the efficiency of traffic estimation with incorporation of weather and holiday data. Such prediction processes can be used in driving simulators for analysis of vehicle dynamics according to different road surfaces condition (wet/dry) due to rainy/snowy weather. Such predictions can also be used in real time traffic management systems for simulating urban traffic with the effect of holidays or special events, for reducing congestion.