Mixed Multi-Unit Combinatorial Auctions (MMUCAs) offer a high potential to be employed for the automated assembly of supply chains of agents. However, little is known about the factors making a winner determination problem (WDP) instance hard to solve. In this paper we empirically study the hardness of MMUCAs: (i) to build a model that predicts the time required to solve a WDP instance (because time can be an important constraint during an auction-based negotiation); and (ii) to assess the factors that make a WDP instance hard to solve. Key words: mixed multi-unit combinatorial auction, machine learning
Pablo Almajano, Jesús Cerquides, Juan A. Ro