Data-driven models for itinerary preferences of air travelers and application for dynamic pricing optimization

Abstract

There is an increasing interest within the travel industry in better understanding customer behavior, particularly the way customers choose between itinerary alternatives when searching for flights. Such an understanding can help travel providers (e.g., airlines) adapt better to market conditions and customer needs, thus increasing their revenue. In this paper, we deal with the problem of modeling air passenger choice between flight itineraries. We describe a two-stage approach to predict travelers’ choice behavior by combining machine learning and discrete choice-modeling techniques. The applicability of these models is then illustrated by employing them for dynamic pricing optimization. We conduct experiments on a dataset extracted from searches and bookings on several European markets, aiming at assessing both the accuracy of our customer models and the effect of price optimization. The proposed approach seems to be effective on both dimensions: (a) improved accuracy when predicting choice, and (b) increased expected revenue of shopping sessions. The experiments show that 42 percent of actual choices fall within the three highest estimated probabilities among 50 alternatives in each shopping session. Moreover, the results also show more than 20 per cent of additional revenue compared with a baseline approach.