A stochastic dynamic pricing model for the multiclass problems in the airline industry

• We propose a dynamic pricing model with multiple products with different behaviors applied to the airline industry.
• We estimate the transition probabilities using the inter-arrival time distribution of bookings and the probability that a booking turns into a ticket.
• We fit these probabilities with phase type distributions.
• Applying this model to a case study improved the expected revenue by 31 percent.
• The phase type distributions fitted with a low error to the data of the industry.

In the airline industry, deciding the ticket price for each flight directly affects the number of people that in the future will try to buy a ticket. Depending on the willingness-to-pay of the customers the flight might take off with empty seats or seats sold at a lower price. Therefore, based on the behavior of the customers, a price must be fixed for each type of product in each period. We propose a stochastic dynamic pricing model to solve this problem, applying phase type distributions and renewal processes to model the inter-arrival time between two customers that book a ticket and the probability that a customer buys a ticket. We test this model in a real-world case where as a result the revenue is increased on average by 31 percent.