Constructing Transit Origin–Destination Tables from Fragmented Data

This study proposes an approach that constructs the origin–destination table (O-D table) for urban bus or light rail lines from fragmented data about the number of boarding and alighting passengers at stops (B-A data) and the analyst's spot knowledge about the trip pattern for selected O-D pairs. The B-A data of transit lines in the city center are often incomplete, yet they may be the only data available to characterize the passenger travel pattern. The proposed approach constructs the O-D table by using data that contain different levels of uncertainties and incompleteness. The model is based on two basic principles, maximum uncertainty and minimum uncertainty. The former is implemented by maximizing the entropy of the O-D table to derive the least-biased values. The latter refers to the maximum consistency with the available data including language-based knowledge about some of the O-D table elements. These principles are implemented by the multiobjective optimization structure. The model is found to be robust if it can incorporate various types of available data as well as the analyst's knowledge. It was tested by using O-D data and B-A data from an actual transit operation. The quality of the derived O-D table is clearly related to the availability and the quality of the data; however, it can be improved significantly with the analyst's spot knowledge about the values of selected O-D pairs. This method will open the way for transit planners to quickly develop a reasonable O-D table from the available incomplete data.