Professor Yusuke Narita, Yale University

Approximating Choice Data by Discrete Choice Models (co-authored by H. Chen & K. Saito)
Tuesday 18 January, 1pm - 2.15pm
Zoom online seminar

Register here

Abstract

We obtain a condition under which mixed logit models are flexible enough to approximate arbitrarily well any stochastic choice generated by the nonparametric random utility models. Our condition is necessary and sufficient and also easy to check. When the condition is not satisfied (and hence, there are some stochastic choice data generated by random utility models that cannot be approximated), we measure the approximation errors and find that the size of the errors is large.

Biography

Yusuke Narita is an Assistant Professor at Yale University. His research involves the data-driven design of decision-making algorithms and mechanisms, with applications to education and health policy issues. His work uses a variety of methods such as causal inference, machine learning, economic theory, and structural econometric modelling. His research has been published in journals including Econometrica, American Economic Review, Management Science, PNAS, NeurIPS (Neural Information Processing Systems), and AAAI (Association for the Advancement of Artificial Intelligence). He obtained a PhD from MIT and was previously a Visiting Assistant Professor at Stanford University.


Further information: business-events@glasgow.ac.uk 

First published: 23 December 2021

<< 2022