Summary: | 碩士 === 國立屏東大學 === 財務金融學系碩士班 === 106 === With fast accumulation of various types of data and the development in new algorithm and cloud operation, big data analysis has been the focused attention by academics and industries. The technologies for data mining has been widely-documented these years. This study apply five data mining algorithm: Decision Trees, Random Forests, Neural Networks, Support Vector Machines and Deep Learning, using RapidMiner to predict credit card holder default risk for a bank in Taiwan. Optimized selection of variables and parameters and cross validation methodologies are used. The result shows that the accuracy rate are all over 80% with Neural Networks displays the best performance. Finally, Neural Networks is used to predict the default risk for randomly selected credit card holders. The outcome can provide valuable information for banks when evaluating customers.
|