Development of Financial Crisis Prediction Models by Using Genetic Algorithm and Neural Network

• Main Author: 邱垂傑
• Other Authors: 周智勳
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/82120464493578732925

碩士 === 中華大學 === 資訊工程學系(所) === 96 === The subject of this thesis is to develop a financial crisis prediction model which contains two main products; the selection of financial feature variables and to establish of crisis prediction model. In the feature variables selection stage, the Genetic Algorithms with two different fitness functions, one is the Fuzzy C-Mean algorithm and the other is a Variance Index, were applied. In the prediction model establish stage, two predictors include Neural Network and Fuzzy C-Mean cluster were used. In the experiment, 125 full-catch delivery companies the crisis samples determined by the static exchange center and 375 normal companies were used as the experiment data. The experimental results concluded that the applied Genetic Algorithms with Variance Index fitness function incorporated with the Fuzzy C-Mean classifier achieved better results than the others.