Center for Machine Learning and Intelligent Systems
About  Citation Policy  Donate a Data Set  Contact

Repository Web            Google
View ALL Data Sets

Shill Bidding Dataset Data Set
Download: Data Folder, Data Set Description

Abstract: We scraped a large number of eBay auctions of a popular product. After preprocessing the auction data, we build the SB dataset. The goal is to share the labelled SB dataset with the researchers.

Data Set Characteristics:  


Number of Instances:




Attribute Characteristics:


Number of Attributes:


Date Donated


Associated Tasks:

Classification, Clustering

Missing Values?


Number of Web Hits:



Ahmad Alzahrani and Samira Sadaoui
alzah234 '@' and sadaouis '@'
Department of Computer Science
University of Regina
Regina, SK, CANADA, S4S 0A2

Data Set Information:

Provide all relevant information about your data set.

Attribute Information:

Record ID: Unique identifier of a record in the dataset.
Auction ID: Unique identifier of an auction.
Bidder ID: Unique identifier of a bidder.
Bidder Tendency: A shill bidder participates exclusively in auctions of few sellers rather than a diversified lot. This is a collusive act involving the fraudulent seller and an accomplice.
Bidding Ratio: A shill bidder participates more frequently to raise the auction price and attract higher bids from legitimate participants.
Successive Outbidding: A shill bidder successively outbids himself even though he is the current winner to increase the price gradually with small consecutive increments.
Last Bidding: A shill bidder becomes inactive at the last stage of the auction (more than 90\% of the auction duration) to avoid winning the auction.
Auction Bids: Auctions with SB activities tend to have a much higher number of bids than the average of bids in concurrent auctions.
Auction Starting Price: a shill bidder usually offers a small starting price to attract legitimate bidders into the auction.
Early Bidding: A shill bidder tends to bid pretty early in the auction (less than 25\% of the auction duration) to get the attention of auction users.
Winning Ratio: A shill bidder competes in many auctions but hardly wins any auctions.
Auction Duration: How long an auction lasted.
Class: 0 for normal behaviour bidding; 1 for otherwise.

Relevant Papers:

Paper 1: Scraping and Preprocessing Commercial Auction Data for Fraud Classification
Paper 2: Clustering and Labeling Auction Fraud Data

Citation Request:

Alzahrani A, Sadaoui S. Scraping and preprocessing commercial auction data for fraud classification. arXiv preprint [Web Link]. 2018 Jun 2.
Alzahrani A, Sadaoui S. Clustering and labeling auction fraud data. InData Management, Analytics and Innovation 2020 (pp. 269-283). Springer, Singapore.

Supported By:

 In Collaboration With:

About  ||  Citation Policy  ||  Donation Policy  ||  Contact  ||  CML