Selection of Support Vector Machines based classifiers for credit risk domain
作者：
Highlights：
• We propose an approach combining feature selection, classification and sliding window testing.
• We propose a technique based on linear Support Vector Machines and particle swarm optimization.
• Proposed technique is also focused on imbalanced learning.
• Proposed technique can be useful to identify “risky” companies which are represented as “minority” in imbalanced datasets.
• Experiment is performed on real world financial dataset from SEC EDGAR database.
摘要
•We propose an approach combining feature selection, classification and sliding window testing.•We propose a technique based on linear Support Vector Machines and particle swarm optimization.•Proposed technique is also focused on imbalanced learning.•Proposed technique can be useful to identify “risky” companies which are represented as “minority” in imbalanced datasets.•Experiment is performed on real world financial dataset from SEC EDGAR database.
论文关键词：Support Vector Machines,SVM,Particle swarm optimization,Credit risk,Default assessment,Classification
论文评审过程：Available online 10 December 2014.
论文官网地址：https://doi.org/10.1016/j.eswa.2014.12.001
