Support Vector Machines with Evolutionary Feature Selection for Default Prediction

Härdle, Wolfgang Karl;Prastyo, Dedy Dwi;Hafner, Christian
(2013) , 24 pages

Files

ISBADP2013-40.pdf
  • Open Access
  • Adobe PDF
  • 988.71 KB
Download

Details

Authors
  • Härdle, Wolfgang KarlHumboldt-Universitat
    Author
  • Prastyo, Dedy DwiHumboldt-Universitat
    Author
  • Hafner, ChristianUCLouvain
    Author
Abstract
This chapter describes a Support Vector Machine (SVM) for default prediction that features evolutionary model selection. It explains that a genetic algorithm was used as an evolutionary algorithm to optimize the SVM parameters and discusses the importance of discriminative power of classification methods in the quality of default prediction. It reviews the support vector methodology in classification, focusing on classical linear and nonlinear classification for linearly separable and nonseparable scenarios. This chapter also evaluates the application of the SVM on the CreditReform database consisting of 20, 000 solvent and 1000 insolvent German companies in the period from 1996 to 2002.
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Härdle, W. K., Prastyo, D. D., & Hafner, C. (2013). Support Vector Machines with Evolutionary Feature Selection for Default Prediction (ISBA Discussion Paper 2013/40). https://hdl.handle.net/2078.5/202629