Cost-Sensitive classification based on decision-theoretic rough set model

  • Authors:

  • Huaxiong Li;Xianzhong Zhou;Jiabao Zhao;Bing Huang

  • Affiliations:

  • School of Management and Engineering, Nanjing University, Nanjing, Jiangsu, P.R. China,State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing, Jiangsu, P.R. China;School of Management and Engineering, Nanjing University, Nanjing, Jiangsu, P.R. China;School of Management and Engineering, Nanjing University, Nanjing, Jiangsu, P.R. China;School of Information Science, Nanjing Audit University, Nanjing, Jiangsu, P.R. China

  • Venue:
  • RSKT'12 Proceedings of the 7th international conference on Rough Sets and Knowledge Technology
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

A framework of cost-sensitive classification based on decision-theoretic rough set model is proposed to determine the local minimum total cost classification and the local optimal test attributes set. Based on the proposed classification strategy, a cost-sensitive classification algorithm CSDTRS is presented. CSDTRS focuses on searching for an optimal test attributes set with minimum total cost including both misclassification cost and test cost, and then determine the classification based on the optimal test attributes set. A heuristic function for evaluating the attribute is presented to determine which attribute should be added in the optimal test attributes set. Experiments on four UCI data sets are performed to show the effectiveness of the proposed classification algorithm.