| Summary: | 碩士 === 朝陽科技大學 === 保險金融管理系碩士班 === 92 === The financial structure is the significant target for National Health Insurance in Taiwan, especially the medical expenditure. This paper applies data mining technology to build the prediction model. We compared logistic regression with neural network. The research results indicate that type I error produced by logistic regression will be changed with cut off value. But, when use misclassification cost to be the evaluate basis, type I error produced by neural network is relatively stable. According to the sensitivity analysis, we find the administrative examination is the most important factor that affects the probability of the professional examination. In other words, an overall checkup on administrative examination can greatly reduce the load of the professional examination. Hence, we can improve the level of auditing.
|
|---|
