Confidence Intervals for OD Demand Estimation

Representative origin-destination (OD) demand tables are a crucial part of making many transportation models relevant to practice. However estimating these OD tables is a challenging problem, even more so determining the confidence intervals on these OD estimates. In this work we propose a method to construct estimates and confidence intervals of OD demand tables from link flow data. Our method separates the uncertainty in estimating OD tables into the statistical uncertainty of link flow data and the possibility of multiple feasible OD demand solutions for the same link flow data. A confidence interval is constructed from concise representations of the uncertainty in each part. We illustrate our estimation method through examples, including one with real data from an intersection in the Los Angeles freeway system.