Using deep learning models to reduce crew standby costs in airline operations
Airlines factor standby crew in their planning as a buffer in case unexpected events cause deviations to operational plans, such as disruptions or last-minute crew roster changes. In these situations, standby buffering is a critical part of crew planning to safeguard the passenger experience and brand reputation. Additionally, the cost of reassigning crews to flights is significantly higher without standby buffers.
Balancing crew standbys with actual utilization on the day of operations is challenging. Our analysis indicates that, on average, only 30%-40% of crew standbys are used, representing a high cost to airline operations. A further look into the figures estimates that, on average, crew costs account for 8.8% of industry costs, and this figure rises to 24% when a disruption strikes. Clearly, there is much margin for improvement.
The good news is that there is a way to lower crew standby costs. By correlating past data from crew rosters with the number of departures per day of operation using machine learning, airlines can achieve 85% accuracy in crew standby forecasts. Moreover, this baseline prediction can further increase by integrating additional analytics airlines may have on actual standby utilization.
Various frameworks can be applied using different deep learning models; each airline will have its own specifics and methodology. But the end objective is the same - decreasing crew standby costs through more accurate forecasts. During the recent AGIFORS conference, my colleague Priyod Chandran explained how deep learning models can help airlines achieve these significant cost savings.
For further information on crew pairing, tracking, and planning optimization, you can also check out the "How a new approach to airline crew tracking helps optimize costs" blog.
Kanu Aravind is the AVP and Head of Digital Innovation unit at IBS Software. He actively works with some of the biggest airlines, hotels, online travel agents, and cruise lines on their digital transformation initiatives. Kanu´s interest in technology ranges from machine learning, natural language processing, and intelligent automation to cloud services and IoT. He regularly advises IBS customers on emerging technologies and their role in digital transformation. Kanu holds a Master's degree from the Indian Institute of Management and is based out of Kochi, India.