Data-driven estimation in equilibrium using inverse optimization

摘要: Equilibrium modeling is common in a variety of fields such as game theory and transportation science. The inputs for these models, however, are often difficult to estimate, while their outputs, i.e., the equilibria they meant describe, directly observable. By combining ideas from inverse optimization with variational inequalities, we develop an efficient, data-driven technique estimating parameters models observed equilibria. We use this estimate utility functions players actions congestion function on road network traffic count data. A distinguishing feature our approach that it supports both parametric nonparametric estimation by leveraging statistical learning (kernel methods regularization operators). In computational experiments involving Nash Wardrop setting, find a) effectively unknown demand or function, respectively, b) proposed substantially improves out-of-sample performance estimators.