We had the pleasure of hosting Ricardo Daziano, Associate Professor at Cornell University, for an insightful seminar on Neural Networks for Choice Analysis: Enhancing Behavioral Regularity and Interpretability.

Professor Daziano discussed innovative approaches to integrating machine learning with econometric principles, focusing on improving behavioral consistency in travel demand forecasting and incorporating network effects into discrete choice models using Graph Neural Networks (GNNs).

Below you can find the abstract and more information about Ricardo’s research interests.

Abstract: This talk explores the integration of machine learning with econometric principles, focusing on behavioral regularization in travel demand modeling. Two case studies will illustrate the need for combining behavioral regularization with advanced machine learning techniques to ensure that models are both theoretically sound and practically useful in travel decision-making contexts. The first case study examines monotonicity in travel demand forecasts from deep neural networks (DNNs). Despite the predictive strength of DNNs, predictions can fail to maintain behavioral consistency. New metrics for assessing monotonicity will be discussed, alongside a constrained optimization framework that exploits gradient regularizers. Applied to travel survey data from Chicago and London, the proposed approach improves DNN behavioral regularity while preserving predictive accuracy, with notable enhancements in smaller samples and out-of-domain scenarios. The second case study introduces a novel graph neural network (GNN) architecture designed to incorporate network effects into discrete choice models, while keeping parametric interpretability. Traditional discrete choice models typically overlook or neglect peer influence, but the proposed GNN approach effectively integrates peer effects, resulting in superior predictive performance and greater interpretability. When applied to New York City commuting data, the GNN model provides valuable economic insights, such as the value of time (VOT), demonstrating its effectiveness in capturing network effects and offering robust policy implications.

Bio: Ricardo A. Daziano, PhD in Economics and Associate Professor of Civil and Environmental Engineering at Cornell University, is a choice modeler working on deriving Bayesian and semiparametric estimators for microeconometric models of customer decisions in engineering contexts. Daziano’s research has a unique focus on applied econometrics of consumer choices around energy efficiency. His technical goal is to derive and apply preference estimators that are econometrically, behaviorally, and computationally superior to extant tools. His empirical goal is to better understand the interaction of consumer behavior and engineering, investment, and policy choices for energy-efficient technologies and sustainability, especially in transportation. In his transdisciplinary career, Daziano has been particularly interested in modeling the socio-technical transition to electric vehicles (EVs), and has used non-market valuation techniques to ascertain preferences for EV features, as well as preferences for EV charging and electricity contracts.