Mode choice in strategic freight transportation models A constrained Box-Cox meta-heuristic for multivariate utility functions (preliminary version)

Modal choice models used in the context of strategic freight transportation studies are generally difficult to set up because of lack of explanatory data. Costs, transit times and trip lengths are often the only figures that can be gathered. While cost and transit time are known as being very important in mode choice decisions, the estimators computed for these variables used in a logit model can be nonsignificant or even have unexpected signs. In this context, Box-Cox transformations of the independent variables can help. Used to improve the log-likelihood of the estimated model, this technique also has the side effect to correct the unexpected signs of the estimators. However, the estimation of the transform parameters can be tricky. Mixing econometrics and operational research, this paper presents a specific shotgun hill climbing meta-heuristic with backtracking capabilities, able to quickly identify Box-Cox  parameters to use when multiple variables must be transformed.