Academic Journals Database
Disseminating quality controlled scientific knowledge

Financial Distress Prediction Based on Cost Sensitive Learning

ADD TO MY LIST
 
Author(s): Wang Hong-Bao | Wang Fu-Sheng | Yang Xian-Fei

Journal: Information Technology Journal
ISSN 1812-5638

Volume: 11;
Issue: 2;
Start page: 294;
Date: 2012;
VIEW PDF   PDF DOWNLOAD PDF   Download PDF Original page

Keywords: Financial distress prediction | cost sensitive learning | support vector machine | cost of prediction error | cost-sensitive support vector machine

ABSTRACT
When companies FDP models have prediction errors, company owners and investors will suffer a great economic loss. Most researches in FDP take the prediction accuracy as the only standard of assessing the quality of financial distress prediction models (FDP). However, these researches ignore the different economic losses of market participants which result from the different costs between type I error and type II error of the model. Therefore, this study proposed companies FDP model based on Cost-Sensitive Support Vector Machine (CS-SVM). The model was established from the perspective of minimizing the cost of prediction error so that it could reduce the loss of users of the model. An empirical research was carried out, taking 86 Chinese listed companies as sample data and adopting 8 times of random sampling to assess the cost of prediction error and prediction accuracy. The result demonstrated that the total cost of prediction error of FDP model based on CS-SVM was only 15.59 which was markedly less than that based on SVM, 23.07.

Tango Rapperswil
Tango Rapperswil

     Affiliate Program