This paper presents an empirical method to predict the airport configuration. We develop both deterministic and probabilistic prediction models. The models do not suggest a configuration that should be used to maximize the throughput, instead predict a configuration that will be used. We use the airport configurations defined in the Operational Information System (OIS) of the FAA. We train and evaluate the models with operational and weather data from 2009 and 2010. The deterministic model predicts a single configuration given a weather condition and time of the day. The probabilistic model predicts a probability distribution of airport configurations given a weather forecast. The deterministic model has an average success rate of about 84%. The accuracy of the probabilistic model is assessed using two approaches, a ranking based approach and an observed frequency based approach. We found that the top three configuration predictions of the probabilistic model have on average a 94% chance of becoming true. We have developed these models for most of the 35 OEP airports.
[1]
Rex K. Kincaid,et al.
A Runway Configuration Management Model with Marginally Decreasing Transition Capacities
,
2010,
Adv. Oper. Res..
[2]
Christopher Weld,et al.
A Runway Configuration Management (RCM) model with marginally decreasing transition capacities
,
2010,
2010 IEEE Systems and Information Engineering Design Symposium.
[3]
Leo Breiman,et al.
Classification and Regression Trees
,
1984
.
[4]
H. H. Hesselink,et al.
Probabilistic 2-Day Forecast of Runway Use
,
2011
.
[5]
Dou Long,et al.
System Oriented Runway Management: A Research Update
,
2011
.
[6]
Katy Griffin,et al.
Quantitative Analysis of Uncertainty in Airport Surface Operations
,
2009
.
[7]
George Hunter.
Probabilistic forecasting of airport capacity
,
2010,
29th Digital Avionics Systems Conference.