Using space syntax and agent-based approaches for modeling pedestrian volume at the urban scale

Contemporary pedestrian volume models are constructed mainly within the space syntax framework with the help of Multiple Regression Analysis (MRA). Although these models predict the distribution of pedestrian volumes in the network with considerable success, they exhibit difficulties in predicting pedestrian movement in some contexts and in accounting for the combined effect of the street network structure and land-use patterns. In this paper we present an agent-based (AB) pedestrian volume model at the urban scale within the space syntax framework. The model was constructed by incorporating transformed basic components of the MRA-based space syntax model to agents' spatial behavior. The AB and MRA models were implemented in two city centers that differ in their urban growth and morphological characteristics. The suggested AB model demonstrated superiority over the MRA model in predicting pedestrian movement when the correspondence between the street network's structure, land uses and pedestrian movement was relatively low and less consistent. We attribute the superiority of the AB model to its ability to represent the combined effect of street network structure and land-use patterns on the distribution of movement flows in an urban network.