| dc.description.abstract | Online Auction is a very popular way for trading in the internet. Huge business opportunities attract large numbers of individuals and stores. With the increase of seller, more and more competitions exist in sellers. At this time, how to maximize profit becomes a crucial issue in online auction business. However, achieving profit maximization goal is not an easy task, it obliges seller to adopt more easily achieving sales amount maximization goal or end-price maximization goal, result in less profit. In our work, we provide a novel service which can predict the expected profit of commodity before listing it and then suggest seller should sell it or not. This service can avoid seller to use inappropriate selling strategy, reduce seller’s cost and further achieve profit maximization goal. We collect data from eBay and use machine learning algorithm to predict sold probability and end-price of listing and further consider the cost of selling strategy to estimate the expected profit of commodity as the basis of suggestion. Besides, we also solve the bias problem in sold probability estimation task and the sample selection bias problem in end-price prediction task. After correcting the bias, our service can obtain more profit, especially under some situations. Finally, we compare the profit of sales amount maximization goal, end-price maximization goal and our service. Our simulations show that our service can get highest profit, sales amount maximization goal get fewer profit and end-price maximization goal tend to be without sale. | en_US |