Analysis of spatial and temporal characteristics of freight demand

Understanding the demand for freight transportation is a key component of the analysis of the freight transportation system. Knowledge of future demand levels is essential to the decision makers involved in the freight transportation industry to assist in their planning processes. However, characteristics of the freight transportation demand preclude the prediction of future scenarios with certainty. Furthermore, many of the forces that govern the interactions among the system elements and its pertinent variables are unknown. Therefore, a systematic analysis of the freight transportation demand is necessary for the design mathematical models that can be successfully applied to forecast future demand. The main objective of this work is to develop a framework to predict the geographical and temporal distribution of freight flows to assist planning decisions at the operational level.; A shipment is defined as a load to be moved from a certain origin to a certain destination by a given mode at a certain time. Two modeling structures, based on the previous definition, are developed in this work to attain the above mentioned objective. The first model is a spatial-temporal multinomial probit which is based on the assumption that demand is generated in a given location, at a given time, according to a potential function which is a proxy measure of the profitability associated with a particular shipment. A shipment will take place whenever the corresponding potential is maximum. The probability that a shipment occurs is obtained through the random utility maximization paradigm. A second model is proposed to determine the latter probability. This model is based on the assumption that freight is generated according to a counting process formed by service requests from several locations at different points in time. It is assumed that these requests follow a compound Poisson process with interactions in space and time.; Both models were tested on simulated data and applied to a georeferenced data set to estimate probabilities of freight transportation demand generation in an actual context.; The specification of the models is assisted by a geographical information system. Details of the implementation of the geographical information system in this framework and of the model specification are presented.