Research process

Three steps of the systematic research process were designed to construct the prediction rules and models of BISE. Figure 1 outlines the research process. First, this study developed a measurement instrument for data collection and conducted validity and reliability analysis to verify its accuracy and the internal consistency of the measurement items. Second, predictive indicators of BISE were determined. Finally, this study identified the critical prediction attributes and indicators and constructed prediction models and rules of BISE using data mining techniques and multivariate statistical methods. Logistic regression analysis and decision tree algorithm are two typical useful methods of data classification and forecasting. The research results will provide practitioners a set of self-evaluation guidelines for BISE estimation.

Fig. 1


Research process


The research focused on the Taiwanese financial services industry. Owing to the global financial services industry being impacted by the Basel II Accord and Sarbanes-Qxley Act, further promotion is needed of the application of new information technologies. For example, banks should comply with financial regulations, thus increasing demand for analytical tools such as intelligence systems and performance management systems.