Modeling pedestrian behavior and decision making has dominantly relied on rational choice models, especially discrete choice models. These models, however, may not be appropriate for modeling the decision processes because they assume unrealistic cognitive and computational abilities of decision makers. Actually, people often use simplifying strategies or decision heuristics for complex decision problems in environments such as shopping streets. The theory of bounded rationality provides a more realistic basis for modeling these kinds of decision problems. Although models of bounded rationality have been studied for a long time, there are no applications in pedestrian research. Most research on bounded rationality has focused on the non-compensatory nature of the decision process or on the characteristics of strategy selection behavior, while other aspects such as factor selection, the coexistence of multiple strategies and integrated frameworks for these aspects have not received much attention yet. This paper therefore proposes such an integrated modeling framework which simultaneously deals with eliciting heterogeneous decision heuristics and the choice of decision heuristic. In addition, we propose an empirical framework for modeling pedestrian decisions including the go-home decision, direction choice decision, rest decision, and store patronage decision. The bounded rationality model is applied to these choice facets. For illustration, a dataset about pedestrian shopping behavior was collected in a shopping street in Shanghai, China. The model estimation results include pedestrians’ preference structures and the probabilities of information search under different search sequences. Estimated parameters are reasonable, providing evidence of the validity of the proposed model. Results indicate that pedestrians tend to use simplified strategies for making satisficing decisions, and tend to search more information for comparative decisions. The predictive validity of the framework is tested using multi-agent simulation. The simulated aggregated spatio-temporal distributions of agent behavior are compared with observed pedestrian behavior. The results are satisfactory in general, however, with some under- and overestimation. Possible reasons are discussed.
|Title of host publication||Pedestrian Behavior Models, Data Collection and Applications|
|Place of Publication||Bingley|
|Publisher||Emerald Group Publishing Ltd.|
|Number of pages||360|
|Publication status||Published - 2009|