A machine-learning approach to a mobility policy proposal.

Area of Science:

• Urban Planning
• Environmental Science
• Computer Science

Background:

• European cities face complex challenges in decision-making for urban development and policy implementation.
• Existing methods for policy testing are time-consuming, hindering efficient urban management.
• The need for data-driven tools to support urban policy analysis and decision-making is critical.

Purpose of the Study:

• To design an open-data, open-source smart-city framework to improve urban decision-making processes.
• To develop a simulation tool with machine learning capabilities for analyzing urban scenarios and policies.
• To accelerate the policy testing cycle for city officials, enabling faster and more effective policy validation.

Main Methods:

• Development of a smart-city framework integrating a simulation tool and a multi-output machine learning unit.
• Deployment of the framework to analyze potential city scenarios, key performance indicators, and utility functions.
• Evaluation of the system using data from Bilbao's Moyua area to test traffic restriction policies.

Main Results:

• The machine learning unit significantly reduced policy verification time from 3 hours to approximately 10 seconds.
• Evaluated strategies suggested potential emission reductions of over 5% for nitrogen oxides (NOx) and particulate matter (PM).
• The system achieved a machine learning accuracy of 91% in identifying effective traffic restriction policies.

Conclusions:

• The URBANITE framework provides a robust and user-friendly tool for enhancing urban policy analysis and decision-making.
• The integration of machine learning accelerates policy testing and provides valuable insights for urban management.
• The framework demonstrates potential for significant emission reductions and optimized traffic management in European cities.