APS Seminar: Applications of Machine Learning to Infrorm Policy-Making for Urban Sustainability

The Department of Asian and Policy Studies (APS) cordially invites you to attend the seminar entitled 「Applications of Machine Learning to Inform Policy-Making for Urban Sustainability」.

Abstract:

Machine Learning (ML) has become ubiquitous in many academic fields. Thanks to their algorithmic nature, ML models are able to capture highly complex and nonlinear patterns that are often not detected by traditional statistical techniques. Nonetheless, a plethora of ML models exist, many of which are often seen as black boxes, and their application to inform policies has been limited to date. In this talk, I will give a brief introduction to ML in general—to explain their superiority to traditional statistics—and I will highlight some efforts that have focus on making ML models more interpretable, including by explaining the concept of feature importance, partial dependence, and SHAP. I will then specifically explain several techniques that have proven useful to model building-scale and household-level energy and resource consumption, and that include probabilistic neural networks, gradient boosting, and probabilistic graph models. Finally, I will apply some of these techniques to travel demand modeling, water consumption modeling, and to determine the interrelationships between electricity, gas, and water consumption. In particular, we will see that ML is able to capture highly nonlinear relationships that makes it an incredibly powerful toolkit to help inform better policies in the future, notably to make cities more sustainable.

About the speaker:

Sybil Derrible is an Associate Professor of Sustainable Infrastructure Systems in Civil and Materials Engineering and Computer Science, a Research Associate Professor at the Institute for Environmental Science and Policy, and the Director of the Complex and Sustainable Urban Networks (CSUN) Laboratory (csun.uic.edu) at the University of Illinois at Chicago. His research is at the nexus of urban metabolism, infrastructure planning and design, complexity science, and data science to redefine how cities are planned, designed, and operated for smart, sustainable, and resilient urban systems. Professor Derrible received a US National Science Foundation CAREER Award for his work and he serves as an Associate Editor for the ASCE Journal of Infrastructure Systems. He is also the Chair of the Sustainable Urban Systems (SUS), Section of the International Society for Industrial Ecology (ISIE) and the Vie-Chair of the Critical Transportation Infrastructure Protection committee of the Transportation Research Board (TRB). He obtained an M.Eng. from Imperial College London and a PhD from the University of Toronto.