Maximum likelihood estimation for trip distribution models

In transportation planning research, the major emphasis in recent years has been laid on disaggregate behavioural traffic demand models. Nevertheless, there is also a strong and continued need for further research in the area of conventional aggregate model techniques. Herein, from a mathematical point of view, the basic models of traffic distribution, namely the gravity and the entropy models, are most fascinating. Conventional solution techniques for these models do not utilize the full power of the modern statistical concepts. The author has undertaken the task to develop and calibrate a comprehensive statistical approach for the estimation of model parameters within the concept and framework of the gravity and the entropy models. He has applied the principles of the maximum likelihood technique to a degree that it can be used in practice. By applying the new technique to a considerable number of real world cases of traffic data for selected cities and regions, he has demonstrated its feasibility. Interpretations indicate, however, that the quality of the achievable results can not be improved beyond the maximum allowed by fundamental (underlying) gravity and entropy models. Also, the quality of the available data has, of course, a limiting effect on the performance of the modelling technique. in summary, the research has created a new and practically applicable traffic demand modelling technique for the step of traffic distribution based on a sound statistical theory. (orig.)Summary in GermanSIGLEAvailable from TIB Hannover: RN 9428(20) / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekDEGerman