The Study of Credit Rating Model For Construction Company

• Main Authors: Yi-Che Sung, 宋宜哲
• Other Authors: Chin-Huang Wang
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/91230580633013431683

碩士 === 國立臺灣科技大學 === 營建工程系 === 88 === This study adopted Artificial Neural Network and financial ratio to build a credit rating model, which can be applied on domestic construction industry. The research samples focused on the construction company from 1994 to 1998, and the financial data is provided by Joint Credit Information Center（JCIC）. We first sifted out twelve significant financial ratios as the model criteria by statistical process. Then the Self-Organizing Map（SOM）neural network was build to cluster the samples into several groups of different credit rating. The supervised outputs were available since the rating of samples were defined. Therefore, the implement of Back-Propagation（BP）neural network was adopted to establish both credit rating classification and prediction model. The training and testing correction rate were 96.29﹪and 87.27﹪for classification model, and 84.37﹪and 62.50﹪for prediction model individually.