Abstract This paper discusses theoretically the statistical tests and measures that are commonly used to evaluate the accuracy of the calibration of gravity models. It is shown that an inappropriate treatment of trip matrices as a certain type of contingency table has yielded erroneous interpretations. Also one of the most commonly used techniques in this context, the correlation coefficient, is inappropriate. Finally a further method of analysis involving the likelihood is proposed.
Statistical notes on the evaluation of calibrated gravity models
1976-01-01
3 pages
Article (Journal)
Electronic Resource
English
Ex Post Evaluation of Calibrated Simulation Models of Significantly Different Future Systems
Online Contents | 2010
|Ex Post Evaluation of Calibrated Simulation Models of Significantly Different Future Systems
Transportation Research Record | 2010
|How Should Travel Demand and Supply Models Be Jointly Calibrated?
Transportation Research Record | 2018
|