ABSTRACT
This chapter proposes a general assessment protocol to validate a choice model against real data and to compare its effectiveness with other models, not necessarily specified through the same approach. It describes a comparison/validation protocol, examples are included for explanatory reasons only. The chapter introduces several benchmarking indicators to validate and compare choice models within a general protocol, proposed for adoption as common practice. Importantly, the final selection, among all the calibrated models, should be made by also taking account of efficiency, or rather, computational speed and memory requirements. The use of multi-criteria analysis techniques based on benchmarking indicators appears worthy of further research effort and will be addressed in future work. It also report results regarding other types of choice models, such as non-linear utility functions through Box-Cox transformations, and fuzzy utility models. Others issues appear worthy of further research effort, such as the definition of the calibration vs hold-out sample.
