Abstract

The collection of origin-destination data for a city is an important but often costly task. This way, there is a need for better and inexpensive methods of collecting information of citizens’ travel patterns. In this line, this paper presents a model to infer the zones of origin and destination for an observed trip between two public transport detentions (i.e., bus stops or metro stations) using socio-economic, land use and public transport information. Therefore, the model enables the reconstruction of the origin-destination matrix of a city, given available trip information passively collected by a smart-card payment system. In order to calibrate the model for Santiago de Chile, a survey was conducted in bus stops and metro stations, with more than a thousand interviewed passengers. We propose a disaggregated zonal inference Logit model including size variables, based on a gravitational model. Regarding our results, we expect that, ceteris paribus, zones with high population have higher probabilities of being the origin of a morning peak trip. Likewise, commercial, industrial, and business centres are expected to have higher chances of being the destination of a trip during this period. Further research is needed to validate our results and develop an efficient algorithm to apply the model and reconstruct the origin-destination matrix. Lastly, we highlight the numerous policy implications of our research, as the information obtained can be used to predict the impacts of changes in the public transport network attributes, as extending routes, relocating their bus stops, designing new routes or changing the fare structure.