Multi-criteria feature selection on cost-sensitive data with missing values

作者：

Highlights：

• A multi-criteria based evaluation function is proposed for measuring features from different viewpoints.

• A dwindling universe is provided to accelerate the feature selection process.

• A feature selection algorithm is developed on cost-sensitive data with missing values.

• The efficiency and effectiveness of the proposed algorithm are demonstrated on different data sets.

摘要

Highlights•A multi-criteria based evaluation function is proposed for measuring features from different viewpoints.•A dwindling universe is provided to accelerate the feature selection process.•A feature selection algorithm is developed on cost-sensitive data with missing values.•The efficiency and effectiveness of the proposed algorithm are demonstrated on different data sets.

论文关键词：Feature selection,Cost-sensitive data,Multi-criteria,Incomplete data,Rough sets

论文评审过程：Received 28 January 2015, Revised 16 June 2015, Accepted 7 September 2015, Available online 28 September 2015, Version of Record 27 November 2015.

论文官网地址：https://doi.org/10.1016/j.patcog.2015.09.016