Toward maximum-predictive-value classification
作者：
Highlights：
• We propose a method for high predictive value classification.
• The new method uses a nearest-prototype approach tuned by a genetic algorithm.
• We compare the method to several standard classifiers optimized for predictive value.
• All methods are tested using 27 real-world classification problems.
• The proposed method produces effective high-predictive-value classifiers.
摘要
•We propose a method for high predictive value classification.•The new method uses a nearest-prototype approach tuned by a genetic algorithm.•We compare the method to several standard classifiers optimized for predictive value.•All methods are tested using 27 real-world classification problems.•The proposed method produces effective high-predictive-value classifiers.
论文关键词：Classification,Nearest prototype,Precision,Predictive value
论文评审过程：Received 15 January 2014, Revised 12 June 2014, Accepted 15 June 2014, Available online 23 June 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.06.013
