Classification driven detection of opportunistic bids in TAC SCM

Abstract

The main objective of a bidding agent in TAC SCM is to get profitable orders and to get enough orders to keep the production going. There is a delicate balance that the bidding agent needs to maintain while deciding on which specific orders to bid and what bidding price to set. In this highly complex bidding problem with (i) many interdependencies, (ii) multiple information flows, (iii) historical data and knowledge, the bidding agent can bid for a few opportunistic orders at a reasonably higher price, which gives higher profit. In this paper, we use classification to determine opportunistic bids to increase our profit. Our solution is robust and adapts according to the dynamic changes in the market condition and the competition provided by the competing agents. Our results show that classification using our opportunistic approach contributes to a significant percentage of our agent’s profit.