政府的財政收入主要來自向人民或企業課徵租稅，因此稅收是穩固一個國家的主要財源。近年來，由於稅務的申報管道多元且便利化，申報案件快速增加，加上逃漏稅的風氣糜爛。在查核人員不足下，無法逐一查核而改為選案查核。因此如何有效的進行逃漏稅查緝及稽查機關效率的提升是個迫切的問題。本研究以台北市國稅局98年零稅率冒退案件作為分析資料。本研究運用資料探勘技術，以SPSS公司之Clementine 12.0套裝軟體及選擇決策樹演算法中C&R Tree、C5.0及CHAID三種方法進行資料探勘的模型建置，俾利找出最佳的規則及預警模式，以改善選案正確率，提升查核效率。
Government incomes come from the tax; therefore tax is an important financial resource of countries. In recent years, due to tax channels are multiform and convenient, tax cases increase so fast and tax evasion is very serious. since lack of staff, tax agency can’t afford to check each case. Therefore, how to investigate tax evasion and improve the efficiency is a pressing issue. In this study, the data are provided by Taipei National Tax Administration in 2009. This study applied data mining techniques, C&R Tree, C5.0 and CHAID, to find out the optimal patterns and models in the case selection for further intervention and investigation by humans
