Сargo Air: A System for Predicting Cargo and Mail Capacities
- The accuracy of the predicted utilization of aircraft cargo/mail capacities increased by 20%
- The accuracy of predicting the available free cargo quota enhanced to 90%
Project prerequisites
The airline’s costs would be reduced by optimizing the utilization of the cargo/mail quota of passenger flights and increasing the accuracy of predicting load capacities.
Goals
Uniqueness of the project
The system core has been developed and trained in such a way as to ensure that not only static historical data and the load of a flight with cargo are taken into account, but also a schedule in which flights can be shifted to an earlier or later time by several hours and days.
The system reliably responds to the dynamic factors taken into account and automatically adjusts indicators. It can also handle newly added flights for which no historical data are available for training.
Customer value
The system implemented has significantly optimized the payload of flights:
- mathematical models are built for accurate prediction of the cargo-mail quota based on analysis of accumulated data
- quotas for new directions are determined based on machine learning
Customer’s testimonial
The implementation of Cargo Air, an automated system for predicting the cargo capacity of passenger flights at Aeroflot, provided an increase in the accuracy of the predictions for aircraft loading for 6 months by 20% and that for available cargo quota by 90%, – says Kirill Bogdanov, Deputy Director General for Information Technology, Aeroflot PJSC. – This allowed us to significantly optimize the total commercial load of flights over the entire Group.
