Data-driven fuel management at Ryanair

Description

Ryanair is the biggest European airlines with a fleet of 400 Boeing 737 NGs making more than 2,200 flights daily and carrying around 120 millions passengers in 2016. And it’s growing—by 2024 Ryanair estimates it will have 200 million passengers. There are multiple ways the company is supported by machine learning. Marcin Pilarczyk explores the main aspects of fuel management of a modern airline (i.e., how to fly cheaper and therefore offer cheaper tickets) and offers an overview of machine learning methods supporting long-term planning and daily decisions.

There are multiple area where flight operations decisions can be supported by using advanced analytics:

• Fuel usage forecasting: Analytical methods based on the planned network and long-term forecasts based on the number of available aircrafts
• Planning support: Identification of routes where excess fuel is carried on regular basis
• Route outlier detection: Supporting flight operations in daily decisions, for instance by analyzing the best slot times
• Line of flight aircraft assignment: Linear integer programming (an old school technique supporting decision making)
• Aircraft weights: Regression models (A proper weight classification reduces ATC costs.)

Multiple data sources are used to provide insights: transactional data belonging to MS world, quick access recorder data, radar data, weather reports, pilot reports, and multiple static sources. The data is processed with Spark on an on-premises Cloudera cluster, Microsoft SSIS, and R and presented mainly with Tableau.