The stochastic assignment model is one of the assignment models enabling multiple route choice. In contrast to the all-or-nothing and equilibrium (capacity restraint) techniques, it is not very widely used, possibly due to its poor results relative to the much simpler all-or-nothing model. One of the reasons for this unsatisfactory performance might be the insufficient calibration of the unknown parameter(s). This paper deals with the calibration problem of stochastic assignment models. Empirical results are reported referring to the calibration of the model for bicycle trips in the city of Delft.
Abstract