Forecast Unplanned Crew Absence

The aviation industry traditionally relied on subjective methods rather than data-driven approaches for some aspects of crew management. Limited insights into crew absence patterns resulted in inefficient workload management and crew allocation strategies. Specifically, standby crews were assigned based on experience as opposed to a more analytical approach. This led to inefficient resource allocation, where on some days crew controllers would have too few or too many standby crews to utilize. There was also limited insight as to what the drivers behind crew absence were

Through the comprehensive analysis of planned andflown rosters, leveraging machine learning techniques like timeseries analysis, our initiative uncovered intricate insights into absence behavior and its correlations with external factors such asweather conditions and special events. Dedicated teams developed anddeployed models to accurately predict crew absences and identifycomplex patterns within the data.

Integrating machine learning into crew managementled to transformative outcomes. By optimizing crew standby allocation through a data and machine learning approach, substantial reductionsin standby crew utilization were achieved, potentially resulting insignificant annual commercial savings for the airline. The detailed analysis provided comprehensive insights into absence drivers,demographic trends, flight destinations, and operationalinefficiencies, empowering decision-makers with actionable data to enhance crew management strategies and operational efficiency.Furthermore, based on a new understanding of the drivers behind crew absences, a further study of the crew experience was conducted. These uncovered aspects related to workload, fatigue and roster stability among others that impact crew satisfaction.