A method to adjust observed transportation data: application to passenger counts on a transit line

Abstract

Traffic volume is a critical piece of information in traffic engineering and transportation planning. The observed values, however, are usually not precise; and hence, adjustment is necessary for them to satisfy various physical constraints, e.g. inflow = out-flow at every junction, in order to make them presentable. We propose a procedure that adjusts the observed values so that they meet the basic physical principle, such as the conservation of flow, given the subjective information about the characters of the observed values Adjustment is based on fuzzy optimization in which the observed values and basic knowledge about the relationship among the observed volumes are considered fuzzy. Fuzzy linear programming is used to find the adjusted values based on the max-min principle. The method is applied to estimate transit ridership at individual stops.