| Summary: | 碩士 === 元智大學 === 資訊管理學系 === 99 === In recent years, there were many cases for bank failure and financial crisis, the US Subprime Lending makes many banks into bankruptcy and enterprises reduce the staff. Therefore, describing to have an early warning system has already become a hot issue. Recently, a lot of research used artificial intelligence methods to build financial early warning system for failure prediction, ex: Back-Propagation Neural Network and Support Vector Machine.
This objective of this study is to use financial variables with a proposed novel model to integrate K-means with Back-Propagation Neural Network (BPN) and Support Vector Machines (SVM) technique to increase the accuracy of the prediction of bank failure. And then compare with Bayes classifier. The research data are provided by Federal Reserve Bank of Chicago. The data set is arbitrarily split into two subsets: about five sixth of the data is used for a training set (1987 to 1992) and one sixth for a validation set (from 1992 to 2008).
Comparing to other methods, the proposed K-means BPN model outperforms other forecasting methods, it not only can increase the accuracy of the prediction of bank failure, but also provides great information for business owners and investors.
|
|---|
