Planners must sometimes synthesize transit route origin-destination (O-D) matrices with limited data, usually on-off counts and sometimes a small or outdated O-D survey sample. When a small O-D sampleis available, iterative methods such as the biproportional method that begin with the sample as a seed matrix can be used, adjusted to match on-off totals. When only on-off totals are available, the recursive method of Tsygalnitsky has been found to match O-D patterns onsome routes better than others. This method is in fact a special case of the biproportional method using an implicit null seed matrix that contains information on directionality and minimum trip length. It illustrates why the recursive method is inappropriate when there is significant competition between routes, and offers a correction for when on-off data have been aggregated to the segment level. Estimation errors are then compared to help indicate how large the seed sample should be in order to produce a more accurate estimate than anestimate produced with a null seed. (A)
Abstract