Study of computational intelligence model of financial crisis prediction applications

• Main Authors: Ho-Wen Cheng, 鄭賀文
• Other Authors: Pei-Chann Chang
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/02828773492297233028

碩士 === 元智大學 === 資訊管理學系 === 98 === In recent years, there were many cases for domestic and foreign enterprises in going out of business. Especially banks, brokerage firms, and other enterprises were devastated by sub prime lending and financial tsunami, which have a great influence on global financial market. Therefore, dealing with financial crisis and enterprise collapse has already become a hot issue. Recently, a lot of research used artificial intelligence methods to build financial early warning system for failure prediction. This objective of this study is to use financial variables with a proposed novel model to integrate Case Based Reasoning (CBR) with Support Vector Machines (SVM) technique to increase the accuracy of the prediction of business failure. The research data are provided by Taiwan Stock Exchange (TSE) and database of the Taiwan Economic Journal (TEJ). The size of matched sample was 120 firms, which a failed firm was paired with two healthy firms by the same industry, products, capitalization, including 40 failed firms and 80 healthy firms. This proposed CBR-SVM model integrated a data clustering technique with Case Based Reasoning weighted clustering, and was then transformed into a smaller case bases and more accurately respond to the current data under classifying from the inductions by Support Vector Machine models. Thus, a financial early warning system for failure prediction based on historical data and financial indexes can be constructed. This study can be divided into three major steps: First, a step-wise regression (SRA) method is applied to select the most important factors from the set of inputs. Second, a Case Based Reasoning (CBR) clustering method is to divide the case library into smaller cases. Third, establishing a Support Vector Machine (SVM) model and output the forecasting results. Comparing to other methods, the proposed CBR-SVM model outperforms other forecasting methods, it not only can increase the accuracy of the prediction of business failure, but also provides great information for business owners and investors. Keyword: Case-Base Reasoning, Support Vector Machine, Financial early warning system for failure prediction