Distance & Flight Time API endpoint has been significantly enhanced to provide more realistic flight time calculations depending on specific route and aircraft type with the help of machine learning (ML) methods.
This endpoint is designed to return the great circle distance and approximate flight time between airports specified by the customer.
Earlier, this endpoint was returning flight time was based exclusively on the great circle distance between requested airports and on the average historic performance of all flights of this length, regardless of the specific route or aircraft type. This was our Standard model of calculations so far.
Now, in addition to the Standard model, you have an option to get flight time calculate based on a far more advanced ML model. This model provides more accurate estimate of flight time between specified airports by taking into account the statistics of the specific route, as well as of an aircraft type provided by the user.
This may come in very handy when you need a realistic estimate of flight times on certain routes. For example, on routes affected by extended airspace closures dictated by war conflicts, sanction regulations, you name it. Such closures typically result in significant detours, and, as a result, in increased flight times comparing to regular estimations, deeming the latter far less relevant.
![For many months, due to airspace closures, flight TK422 and many others in the region are taking a much longer route to reach their destination. A flight that would take about 2.5 hours, now takes almost 5 hours.](https://aerodatabox.com/wp-content/uploads/2024/07/2024-07-flight-time-ml-TK422.gif)
The Standard model estimates this flight to be completed with 2 hours 10 minutes, which would have been about right in normal circumstances, but not in this case.
The ML model estimates this flight to be completed with 4 hours 50 minutes, which corresponds much better with the average flight time on the route recently. See more examples below.
Another good example where ML model is that on some routes, there could be generally more delays than on other routes of the same distance.
Route | Average Flight Duration, past year | Standard Model Flight Time Prediction | ML Model Flight Time Prediction |
---|---|---|---|
Amsterdam (EHAM) to Istanbul (LTFM) | 03:30 | 02:45 | 03:40 |
Istanbul (LTFM) to Amsterdam (EHAM) | 03:10 | 02:45 | 03:10 |
Los Angeles (KLAX) to New York (KJFK) | 04:55 | 05:00 | 04:50 |
New York (KJFK) to Los Angeles (KLAX) | 05:30 | 05:00 | 05:20 |
Moscow (UUEE) to Istanbul (LTFM) | 05:00 | 02:15 | 05:10 |
Istanbul (LTFM) to Moscow (UUEE) | 04:35 | 02:15 | 04:50 |
Another new thing is that now both ML and Standard models of Flight time calculation are able to attempt to take the aircraft type into account, if specified by the API user (subject to data availability about the type).
We also ensured backwards compatibility. The Distance & Flight Time API endpoint shall continue to return results according to the old, Standard calculation model, so nothing changes for the existing users, unless they decide to switch.
If you want to use the new model, please make sure to include extra parameter as highlighted below.
GET /airports/:codeType/:code/distance-time/:codeTo?flightTimeModel=ML01
If you want to try refining your estimate using aircraft type, please also add the following
GET /airports/:codeType/:code/distance-time/:codeTo?flightTimeModel=ML01&aircraftName=Airbus A320
You can find more information about this endpoint in the documentation.
We hope you will enjoy this new update!