Multiple classification analysis in trip production models

We analyse various Multiple Classification Analysis (MCA) methods to model trip production (generation). We first show that the MCA version most widely used in transport engineering implies a rarely feasible assumption, the transgression of which may drive a significant overestimation of the future number of trips and a systematic bias in its socio-economic composition. To illustrate this effect, we use Monte Carlo simulation and real data from Santiago, Chile to compare the various MCA approaches, concluding that the aforementioned form should be discarded. Our analysis also shows that the MCA method which is more robust to the structure of the underlying model, is the simple calculation of trip rates as averages for each category. Finally, we hint at the need to use more sophisticated formulations than MCA to model trip production.