Data driven design for online industrial auctions

Designing auction parameters for online industrial auctions is a complex problem due to highly heterogeneous items. Currently, auctioneers rely heavily on their experts in design. The ability of predicting how well an will perform prior the start comes handy auctioneers. If item expected be low-performing item, auctioneer can take certain actions influence outcome. For instance, starting selling price modified, or location where displayed website changed attract more attention. In this paper, we real-world data set and investigate improve upon expert’s design using insights learned from data. More specifically, first construct classification model that predicts performance auctions. We propose driven framework (called DDAD) combines knowledge with prediction model, order find best parameter values, i.e., display positions items, given new auction. evaluated, several discussed validated experts.