Flights Delay Prediction

Can Machine Learning help reduce flights delay?

Flight delays cost the American airline industry alone $22 billion yearly. Sometimes multiple delay events are so intertwined that it is impossible for an airport management team to segregate their sources, leaving each stakeholder: security, maintenance crew, fueling crew, technical crew, involved in the generation of the delay without means to anticipate and therefore manage it.

However, since June 2003, the airlines report both on-time data and the causes of delays and cancellations. These existing data can be used to nourish predictive models, which in turn will provide an estimate of future flight delays.

Problems to solve

  • How to predict flight delay?
  • How can we improve the passenger experience in an airport?
  • Can Machine Learning help airport personnel work more efficiently?
  • Can a flight actual arrival/departure time be anticipated?
  • Can the maintenance staff be called in advance based on actual aircraft arrival?
  • Can the airport organisation be adapted to this anticipated delay rather than suffer it?
  • Can the relevant staff in each airport profession be called when really needed rather then when expectedly needed?
  • Can the whole organisation of the airport be reengineered to be based on predicted actual flight departure and arrivals rather than scheduled ones?
  • Can an airport gain significant customer service improvement?
  • Benefits of TADA
    in the aviation industry

    A great variety of professions work in airports:

    • air traffic control similar to police officer directing traffic on a busy street,
    • security such as running the X-Ray machines to monitoring traffic in and out of the airport,
    • baggage handler who lift and transport cargo to the correct destinations,
    • pilots and flight attendants, among others.

    These professionals can use predictive models to predict flight delays and improve their organization. However, they are usually not Data Scientists and rarely have the required skills in Machine Learning nor in software coding to build predictive models.

    Most of the historical data regarding airplane take-off and landing is based on a small number of flights. For instance, the second largest airport in France, Orly supports the take-off and landing of 250 000 flights a year, less than a thousand a day. The type of data collected is dates, actual departure times versus scheduled departure times, airline carrier, time spent by the aircraft in the air, departure and destination, distance, weather. This means that these data are mostly small quantities of data, also known as Small Data. Traditional Machine Learning tools can’t handle Small Data. In this context, MyDataModels offers a solution to provide airport staff with a means to automatically build predictive models out of their Small Data.

    No training is required to use TADA, MyDataModels solution. Airport professionals can use their own data without any kind of data formatting: normalization or pre-processing, or feature engineering. MyDataModels brings a self-service solution for those who have Small Data and no data scientists.

    TADA brings new possibilities for airport personnel

    Flight delays have severe consequences on different areas of an airport organization: track management, luggage delivery, plane maintenance, cleaning staff, security. Flight delays disturb the organization of an airport, and impact the efficiency with which the different airport professions interact, resulting in massive costs for these professionals.

    “By using Machine Learning professionals can easily predict potential delays”

    By using Machine Learning and TADA, these different categories of professionals can easily predict potential delays in a few clicks (and a few minutes), and adapt their assignment and planning of resources accordingly while simultaneously providing passengers with the best customer experience.