Improving departure taxi time predictions using ASDE-X surveillance data

Aircraft taxi time prediction: Feature importance and their implications

“…Notice only data on good weather days were applied to train and test the model. Building on a historical traffic flow database, Srivastava [18] established an adaptive taxi time prediction model with LR analysis, where a set of explanatory variables including aircraft queue position, taxi distance are included. Using actual data from John F. Kennedy International Airport (JFK), the prediction model has been demonstrated with high accuracy.…”

“…Various modelling techniques have been utilised in the literature to estimate and predict the taxi time, e.g., queuing models [12], statistical regression approaches [11], fuzzy rule-based systems [13] and machine learning techniques [14]. Meanwhile, existing research has adopted different numbers of features (from 5 up to 42) that may affect taxi time from different data sources, including Airline Service Quality Performance (ASQP) and Preferential Runway Assignment System (PRAS) [12,15], Aviation System Performance Metrics (ASPM) [14,16,17], Airport Surface Detection Equipment, Model X (ASDE-X) and Severe Weather Avoidance Programs (SWAP) [18,19], Spot and Runway Departure Advisor (SARDA) [20,19], FlightRadar24 (FR24) [21]. This leads to a potential risk that certain features related to the taxi time have not been included.…”

Measuring Flight Efficiency in the National Airspace System

“… Aviation System Performance Metrics (ASPM) 6 : runway information (from CountOps 7 ), plus gate-out, gate-in, wheels-off, and wheels-off times (fused from Out, Off, On, In [OOOI], Airline Service Quality Performance [ASQP], ADL, and TFMS data)  GroundTracker (Srivastava, 2011): terminal and taxiway segment information, derived from raw ASDE-X position data Finally, the taxi efficiency metrics are initially computed for each individual flight in the dataset. Hourly, daily, and weekly efficiency metrics are then computed for the corresponding set of flights for each time period using a time-binning approach.…”

The Distribution Function of Airport Taxi-Out Times and Selected Applications

“…On the other hand, Airport Surface Detection Equipment, Model X (ASDE-X) seems to be the most ambitious hardware/software combination to build an adaptive taxi-out prediction model based on a historical traffic flow database generated by the system itself (Srivastava 2011). It provides high resolution coverage of aircraft surface movement.…”