A Data-Driven Approach for Ancillary Bundle Recommendation to Segmented Users
Aakash Swami, Narendhar Gugulothu, V Tirumala, Sanjay P. Bhat · 2024
Recently, ancillary revenues and targeted offers for diverse users have become increasingly relevant. Consequently, service providers have begun recommending customized bundled ancillary offers to specific users. Ancillary bundle recommendation involves selling one or more ancillary products as a single unit and entails pricing and assortment optimization. In this paper, we propose a novel approach for ancillary bundle recommendation using item-level purchase data for a given user segment, defined based on user travel attributes. While many service providers lack bundle-level purchase data, item-level purchase data is readily available. Our proposed approach, Neural Segment-based Ancillary Bundle Recommendation (NSABR) leverages item-level purchase data and involves several key steps: First, in data preparation, we initially convert item-level sales data to bundle-level sales data. Then, for each purchase event, we sample non-purchased ancillary bundles to ensure our dataset includes both positive (purchased) and negative (not purchased) instances. Second, we utilize a novel neural network-based binary classification model to predict the probability of purchase of a given ancillary bundle at a given price for a specific user segment. Third, we price the ancillary bundles for a given user segment by maximizing revenue. Finally, we make the recommendations for a given user segment by ranking ancillary bundles based on their estimated revenue by selecting the top-N bundles. We conduct experiments using synthetic data and compare our proposed approach, NSABR, to true synthetic simulator results and a few baselines to demonstrate its applicability.