Machine-Learning Approach for Mobility Policy Proposal

Abstract

The Urbanite project is an open-data, open-source framework that aims to help European cities make better decisions by providing them with a powerful and easy-to-use simulation tool. The framework includes a machine learning (ML) module that allows city officials to analyse large amounts of data and identify patterns and trends that can inform policy decisions. The ML module uses a multiclass-multioutput approach, which allows for the simultaneous prediction of multiple outcomes based on a wide range of input variables. The main objective of the ML module is to enable decision-makers to define potential city scenarios and a utility function, and the ML model will help find a policy that best applies to the given constraints and preferences. Among essential improvements is the speed-up of several orders of magnitude for policy testing, provided by the ML module. The system was tested in the Moyua area of Bilbao, where it was able to achieve a predefined decrease in emissions and other key performance indicators (KPIs). The system was able to provide useful insights into the best policy for closing certain districts to private traffic and the best times for those closures based on data from simulated and learned scenarios.

Keywords: machine learning, Smart cities, mobility policy