Data and AI
Machine Learning, Digital Transformation
Travel | Hospitality
The client is a leading Dubai (UAE)-based airline offering award-winning services across six continents.
Understanding the Challenge
The airline was using manual flight prioritization processes in case of flight delays due to weather and non-weather-related disruptions. In case of weather disruption, the capacity of the runway reduces to as low as 20%, so departing flights were put on hold on the ground and arriving flights were managed in the air. In case of non-weather disruption, flights were delayed due to reasons like runway closure, mechanical defects, etc. The challenge the airline was facing was manually deciding which flights should be delayed and by how much time. This method left a lot of room for inaccuracy, severely affected the overall customer experience, and had negative cost implications.
Xebia’s Strategy and Solution
Spearheading transformation with machine learning and optimization mechanism for flight prioritization
Xebia charted a complete digital transformation roadmap for replacing the manual flight prioritization system with an automated data-driven decision system powered by the latest technology at its core.
A mechanism that computed the value score of each flight was developed. This value score was calculated using various passenger-related parameters like age, gender, loyalty tier (if any), number of infants, medical case/wheelchairs, business/economy class, connecting flight timings and more.
- The passenger parameters were fetched from the reservation platform system, departure control system, and loyalty program.
- The process followed to derive flight score included feature engineering, computing the optimum number of clusters, creating clusters, validation and prioritization, feature importance, and flight scoring.
- Clustering algorithm was used to segment flights into profiles and each profile was validated with business users to arrive at the priority order.
- Clustering was followed up with scoring of flights within each cluster/profile by computing the optimal weight for each parameter.
- This ensured that all individual flights in a high priority cluster would have a higher score, and hence have higher priority than those in a lower priority cluster.
The implementation of the automated data-driven decision system enabled the airline to significantly improve its operational efficiency through real-time decisions, which contributed to:
- Reduction in airline operating cost
- Improved traffic handling capability
- Improve accuracy in the decision-making process
- Elevated customer experience