Using machine learning to analyze air traffic management actions: Ground delay program case study. (November 2019)

• Record Type: Journal Article
• Authors: Liu, Yulin; Liu, Yi; Hansen, Mark; Pozdnukhov, Alexey; Zhang, Danqing
• Is Part Of: Transportation research. Volume 131(2019)
• Page Start: 80
• Page End: 95
• Publication Date: 2019-11
• Subjects: Ground delay program -- Convective weather -- Support vector machine -- Logistic regression -- Random forest -- Regularized linear models -- Feature importance; Logistics -- Periodicals; Transportation -- Periodicals; 388.011
• DOI: 10.1016/j.tre.2019.09.012
• Languages: English
• ISSNs: 1366-5545

Abstract

Highlights: Propose two-stage machine learning model to predict GDP from high-fidelity weather data. SVM model learns spatial pattern in convective weather that triggers GDP. Weather pattern extends along coastline and exhibits elliptical shapes around airports. Random forest indicates good prediction power, temporal predictability, and transferability. Feature importance suggests weather and demand are dominant factors in predicting GDP. Abstract: We model the impact of weather and arrival demand on ground delay program (GDP) incidence. We use Support Vector Machine (SVM) to analyze how regional convective weather affects GDP incidence and find the impact depends on both distance and direction of convective activity from the airport. We then train and compare the performance of logistic regression (LR) and random forest (RF) in predicting GDP incidence using an SVM-generated regional weather variable, local weather and arrival demand. Generally, RF outperforms LR. Convective weather is the most important factor in predicting GDP incidence at Atlanta International Airport (ATL), while arrival demand has greater impact for the other airports studied. We also examined model transferability across different airports. Lastly, we build GDP duration prediction models to enable a user to assess how long a GDP is likely to continue, if it is in effect in a given hour.