The market for used trading card games (TCG) has rapidly grown in recent years. In the TCG market, the demand for used products tends to increase with the release of new products, thus increasing the costs. We researched the used sales market for “Magic: The Gathering,” one of the TCGs, with the goal of predicting skyrocketing product costs. However, in terms of prediction, there are few emerging products, landing to imbalanced data and a dearth of ”soaring” products. Using current imbalanced data handling techniques, we increased the recall of imbalanced data in this research’s predictions of TCG skyrocketing price. As a result, the imbalanced data handling method enhanced recall that the methods were effective for predicting price increases. Additionally, we tried to compare the methods of oversampling, undersampling, and cost-aware learning approach. It was thus demonstrated that the costaware approach and undersampling were particularly successful handling methods.

Authors: Takeru Shishido, Ayahiko Niimi

Published in: International Conference for Internet Technology and Secured Transactions (ICITST-2022)

  • Date of Conference: 6-8 December 2022
  • DOI: 10.20533/ICITST.2022.0008
  • ISBN: 978-1-913572-55-6
  • Conference Location: Virtual (London, UK)