Learning revenue-maximizing orderings in sequential auctions

When multiple items are auctioned sequentially, the ordering of auctions plays an important role in the total revenue collected by the auctioneer. This is true especially with budget constrained bidders and the presence of complementarities among items. In such sequential auction settings, it is difficult to develop efficient algorithms for finding an optimal sequence of items. However, when historical data are available it is possible to learn good orderings that increase the revenue of the auctioneer. In this work, we show how such a learning model can be built based on previous auctions using regression trees. We provide a greedy method that finds a good sequence for a new set of items given the learned model. We design several experiment scenarios and test the performance of the proposed learning method. The experimental results are promising: they show that good orderings can be found quickly.