User-interface development for Brisa´s newly developed traffic forecast models

The following study summarises the outputs of the 18 months project-based-learning course together with Portuguese motorway concessionaire Brisa. Thereby, the study covers (traffic) forecasting models includings arima and facebook prophet as well as UI design principles. Further, the paper compares the performance of traffic volume forecast models for the 33 different motorways, using different sets of exogenous variables. Finally, the paper describes the development process of a new traffic volume forecast dashboard using Dash Plotly, a Python library. The work concludes, that using different(non-intentional) exogenous variables lead to very good model performance and by developing a user-interface, non-technical users get a chance to interact with (more)complex models.