Analysis of bid history in Internet auction and its application

Recently, Internet technology has progressed, and there are many people who use the Internet. Internet auction are one of the services using Internet technology. Many users of Internet auctions want to get items which are cheaper and/or rare. However, there are many situations where the bid price goes up, especially in the phenomena in which the bid price rises suddenly towards the end of an auction. This leads to many users making bids higher than estimated because they get excited. For this reason, the data of market price-related contract price for Internet auctions was unveiled. However, there was no data for new or rare items. Therefore, it is necessary to support bidding activities in situations involving new or rare items. In this study, we analyzed contract prices using bid histories, not market price. Bid history was classified into seven clusters by clustering based on the k-means method. We verified that each piece of classified data has the features of contract price, opening price and so on. We then propose an estimation method of contract price based on our clustering. As the result, we show that users can judge calmly when provided with the estimation contract price.