본문 바로가기
  • Home

Prediction Performance Improvement of KOSDAQ Business Bankruptcy using SVM Preprocessor

  • Journal of Knowledge Information Technology and Systems
  • Abbr : JKITS
  • 2012, 7(4), pp.21-27
  • Publisher : Korea Knowledge Information Technology Society
  • Research Area : Interdisciplinary Studies > Interdisciplinary Research
  • Published : August 31, 2012

Kang, BooSik 1 Cho, Jun Hee 1

1목원대학교

Accredited

ABSTRACT

Business bankruptcy has had a lot of losses to various stakeholders. If we are able to predict bankruptcy or insolvency in advance, we can defend enterprises against bankruptcy or insolvency and minimize the loss. Various types of statistical and data mining models have been proposed for prediction of business bankruptcy. One of the main issues of the bankruptcy prediction models is to increase predictability. This research deals with logit, decision tree, neural networks, and SVM models those are mainly used to predict corporate bankruptcy. Sample companies were consisted of 49 bankrupt firms and 49 normal firms in KOSDAQ. Nine financial variables were selected by statistical test and five years of data were collected. In case of using SVM for preprocessing data, experimental results showed that classification models could improve the performance of the bankruptcy prediction.

Citation status

* References for papers published after 2023 are currently being built.