Applying Back Propagation Neural Network and Sequential Pattern Mining to Construct Corporation Crisis Prediction Model–A Case of Taiwan’s Electronic Industry

• Main Authors: Li-Jie Hon, 洪立劼
• Other Authors: 羅淑娟
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/89y3ap

碩士 === 國立臺北科技大學 === 工業工程與管理研究所 === 96 === A bankruptcy prediction model is often built upon the information which comes from financial statements. Many researchers adopt statistical methods or artificial intelligence to build the classification model and use the model to predict the future status. Since these models require financial information to judge or predict the operational situation, it is impossible to predict without any financial data. Our research tries to combine Back-propagation Neural Network(BPNN) and sequential pattern mining to overcome this drawback. We use two ways to match our distress and non-distress data by considering industrial factors and use samples from different period to build the classification models. We see classification result from the models as signals, which means distress or non-distress at specific term and furthermore, we mine those signals in order to get some patterns which help us do prediction. We experiment on financial data of Taiwan’s electronic industry from TEJ database and the result shows the combination of BPNN and sequential pattern mining can predict the operational status efficiently.