Exploring Heuristics Underlying Pedestrian Shopping Decision Processes : An application of gene expression programming

Most analytical pedestrian behavior researches use utility-maximizing models and have paid less attention to models based on alternative behavioral theories such as bounded rationality. Consequently, there is a lack of deeper explorations into the decision processes of pedestrians. This lack of such alternative models may also be the result of inappropriate methods to estimate such models. For this reason, the paper first introduces a modeling platform GEPAT which has the GEPAT which has the ability to estimate parallel functions using a multi-gene-sectional chromosome structure and to facilitate building models using processors emulating simple decision mechanisms. The going-home decision of pedestrians in Wang Fujing Street is taken as an example to illustrate the use of GEPAT. The most important conclusion from a comparison of the MNL, hard cut-off, soft cut-off and hybrid model is that the satisficing heuristic fits better to the problem structure, at least in this case, than the utility-maximizing rule does. This example also shows the flexibility of GEPAT as a modeling toolbox and the power of estimating complex models.