Except in adverse weather conditions, congestion at large airport hubs appears to be predictable. This paper attempts to translate this predictability into a distribution of taxi-out times, a key component of airport congestion. When scheduled flights are chosen to define the dataset, taxi-out times follow a uniform distribution. This is not only the simplest distribution that inferences can be based on, but also a distribution that can be estimated by simple linear regression leading to very accurate forecasts. But above all, it is an invertible distribution function that can help solve a large class of stochastic optimization problems.
