A heuristic for incorporating ancillaries into air choice models with personalization (part 2: integrated multinomial logit and hedonic regression models)

Abstract

In recent years, airlines have begun selling “branded fares” which are bundles of a travel “right-to-fly” plus assorted amenities (e.g., first bag included, extra legroom, priority boarding, extra loyalty miles, etc.). Branded fares allow airlines to further differentiate their products and have proven popular with customers. Logit-based discrete choice models are used in a variety of air travel decision support applications; a widely used one is online share estimation of itinerary/fare results returned from air shopping low fare search. However, the explanatory features considered in previous air choice models in use at Sabre were limited to price and itinerary schedule attributes only; they did not consider the value-added utility of the new amenities included in airline branded fares. This paper is the second of a two-part series about incorporating ancillaries into air choice models. In Part 2, the authors describe an approach for integrating multinomial logit and hedonic regression models by customer segment (allowing consideration of the value-added utility derived from bundling ancillary amenities in airline branded fares). When applied to shopping results that include branded fares, we show how this new approach leads to 25% improvements in air choice model prediction accuracy (compared to using price and schedule only features). We will also present a new choice-model accuracy metric (TOPN%) that allows comparison of results across samples with different assortment sizes.