Predicting the success of group buying auctions via classification

Online group buying involves risks and uncertainties for buyers. Because buyers may not benefit from a failed group buying auction, they usually consider several factors before deciding to join an auction. Although the factors that affect buyers’ decisions have been investigated in the literature, few studies have attempted to utilize them to predict an auction’s success. In this paper, we propose an effective method for predicting the success of a group buying auction. We model success prediction as a classification problem, and utilize five dimensions and thirteen features derived from previous research to predict the success of group buying auctions. Experiments based on a real-world group buying platform demonstrate the efficacy of the proposed method. Moreover, the method outperforms a social propagation model in terms of the prediction precision rate, recall rate, and F1 score.