Since the early 1970s several techniques have been developed for estimating origin-destination (o-d) matrices using all kinds of data. This paper presents a survey of four different o-d matrix estimation techniques. It is shown that some techniques can be applied only in certain restricted conditions and, even worse, may lead to considerable loss of information. To alleviate this problem, it is suggested that elastic, instead of fixed, constraints be used. It is also possible to overcome this problem by using an explicit stochastic estimation technique (binary calibration). Because of the absence of time and space-dependent coefficients, this technique can also be used for making forecasts. This paper appears in Transportation Research Record No. 1203, Demand Forecasting and Trip Generation-Route Choice Dynamics.
Abstract