Market Segmentation in Aviation

Challenge

To model the purchasing power of passengers, estimating price elasticities is an important approach for airlines. Our client, an international aviation corporation, faced the challenge of determining optimal price elasticities for a variety of customer segments across different markets.

Approach

One of the biggest challenges in estimating segment-specific elasticities is that the airline's routes have varying frequencies, yet price elasticities need to be estimated for all products. Booking data from past years was utilized and enriched with additional flight-specific information. Instead of manually specifying which segments should have elasticities calculated, the model iteratively determines the best segmentations itself. This results in data-driven market segments, each with its own estimated price elasticity. The estimation itself is achieved through a client-specific modified gradient boosting algorithm. Due to the client's requirement to use the model for as many markets as possible simultaneously, the methodology was further developed for use on a high-performance computing system in the cloud using Spark.

Result

Through this novel methodological approach, the client can now more precisely play out the optimal price in the booking system for all products. The segmentation is handled by a model-based approach to obtain similar segments across different products. Interactive analysis of the segmentation results and estimated elasticities can also be done through dashboards we provided.