Machine-learning opportunities within the airline industry

Rodrigo Fontecilla explains how many of the largest airlines use different classes of machine-learning algorithms to create robust and reusable predictive models to provide a holistic view of operations and provide business value.

Rodrigo Fontecilla explains how many of the largest airlines use different classes of machine-learning algorithms to create robust and reusable predictive models to provide a holistic view of operations and provide business value. Rodrigo offers an overview of how these models support day-to-day operations for the airlines. For instance, a day before flight operations, based on a reasonably accurate weather forecast for the next day, the probability of losing connections in hub airports for the specific airline is estimated. If the probability is above a threshold value, the airline reservation system will be involved to provide the number of possibly affected passengers. If that number is above a predetermined value, the passenger travel recovery system is involved to simulate the re-accommodation of these passengers on later flights. The predictive model provides the potential number of passengers that would need meals and/or hotel rooms or that would be eligible for a compensation (all based on airline-specific and legal rules). Apart from being ready to take care of the affected passengers quicker if lost connections materialize, the airline learns the economic impact of a flight’s disruption and is better prepared to deploy its staff in a timely manner to handle the disruption in the airport. By getting this data in advance, the airline can decide if it’s worth re-accommodating the passengers on another airline. Rodrigo explores the business benefits of predicting lost connections to the airline and the airport, including the possibility of flying the most valuable passengers through other airline hubs if there is a high probability of losing the connection from the original passenger booking.