Inferring complete demand and actual capacity information from the airline origin and destination survey

The methods proposed are intended to estimate historical air travel demand and actual service capacity between origin and destination airports. The results obtained using these methods could prove useful in, for example, evaluating the efficiency, and calibrating demand analysis and forecasting models for the air transportation system. More specifically, we attempt to address two issues in the Airline Origin and Destination Survey data. One is that the sampled travel demand is not always reliable due to sampling errors. The second one is that it contains little information about the actual capacity — net of the capacity absorbed by connecting demand — between origin and destination airports. To address the first issue, we designed a constrained least square model to estimate the complete historical travel demand served by the survey participating carriers. To address the second issue, a novel adaptive adjustment capacity estimation scheme was proposed, and embedded into the solution algorithm of the constrained least square model. We applied the methods to the 2007 survey data. We found that 1) on average, about 75% of the itineraries' actual capacities are utilized and the variance is about 7%; 2) the survey oversampled the total demand by about 0.4% and the variance of sample rate is about 1.1%.