✈️ Flight Performance App

Stephen Bique, Ph.D.

  • Quickly compare airlines using actual elapsed times and delays
  • Instantly obtain expected delay on arrival
  • Highly flexible with full spectrum of options for cities, airline and flight times
  • Fast, responsive, efficient with professionally designed layout
  • User-friendly GUI with familiar click and scroll interface

To test the application, visit Link to Flight Performance App.

Why Do We Need A Model?

Suppose you know the departure delay, but the plane has not yet left the gate. Presumably, the pilot will eventually give the best answer when more information is available. What is a good estimate of the arrival delay as soon as you know the likely departure delay? As our illustration shows, a guess that the arrival delay is the departure delay turns out to yield on average the worst relative error (blue).

Depiction of Retive Error for Different Models A much better estimate is obtained using the best regression line with the departure delay as the independent variable (orange). We introduce a new model which yields on average the smallest relative error (green).

New Model

We propose the following model to estimate the arrival delay:

departure delay + flight time + taxi-in + taxi-out - elapsed

  • departure delay is the difference between scheduled departure and actual push off from gate in minutes (often negative)
  • flight time is an estimate of the time from takeoff until landing
  • taxi-in is an estimate of time from pushing away from the gate to takeoff
  • taxi-off is an estimate of the time from landing until parked at the gate
  • elapsed time is the scheduled duration from gate to gate

We conducted studies to find good estimates for the chosen variables to minimize relative error in the predicted arrival delay.

Flight Performance GUI

Predict Delay on Arrival

Compare Flight Performance of Airlines Online Help and About tabs are available in addition to tabs shown as they appear on a popular mobile device.

Summary of Help Tab

  • Select the origin and destination from the drop-down lists. The locations are listed by the (IATA) airport codes, which may be found using the lookup menu provided.
  • For the Expected Delay tab which displays the expected delay:
    • Select the desired airline from the drop-down menu.
    • Choose any desired departure time. The departure time is the scheduled time the aircraft pushes away from the gate.
    • Select the known or expected delay of departure, which is expandable.
    • Select the elapsed time (if shown), which makes it possible to avoid time zone details.
  • For the Compare Airlines tab, uncheck departure times only to exclude the relevant data for such departures. To toggle the visibility of an item, click on that item in the legend. To toggle the visibility of other items, double click on that item in the legend.

Conclusion

  • We proposed a new model to study airline performance and have shown this model yields the best estimate of the delay on arrival based on relative error.
  • RStudio provides a convenient and efficient software environment to perform all needed database operations. The chosen BTS database is suitable to demonstrate the power of R because the data is in multiple tables and it is necessary to clean the data to handle missing data.
  • The Flight Performance App quickly displays the calculated expected delay on arrival, which is helpful information whenever a flight's departure is delayed.
  • The Flight Performance App quickly provides an interactive plot to compare airlines by elapsed time, which is a better measure of airline performance compared to published on-time performance, and thus useful information to consider before reserving airline tickets.