Applying Back Propagation Neural Network and sequential pattern mining to construct corporation crisis prediction model -A case of Taiwan's electronic industry

Traditional studies utilize data from financial statements to construct financial crisis alerting models by using statistical methods or artificial intelligence techniques. According to these models, researchers can evaluate the possibilities of bankruptcy. While these models can predict the future operational status with the aid of earlier financial data, they cannot consider the continuous classification signals which might be helpful to the prediction. Hence, according to priori studies, our study not only considers corporate factors and different sample matches but also uses financial ratios within different periods to construct Back-Propagation Neural Network models. Besides, we apply sequential pattern mining to signals of the best models, which we construct, in order to gain some prediction patterns, which we can utilize to predict a company's operational status of next term. The study shows that the combination of Back-Propagation Neural Network model (BPNN) and sequential pattern mining can forecast the operational status without any financial information of next term and its predicting result is similar to the classification result which is made by BPNN.