隨著少子化的影響，教育部決定透過新生註冊率作為私立大學經營存續的依據。由於推薦甄試的比例提高，如何在推薦甄選入學過程中，掌握並選擇有強烈就讀可能的學生是各大學特別是私立大專院校最重要的課題。過去本校對於推甄資料的分類都透過人為主觀因素判斷，又缺乏後續追蹤驗證。本論文透過資料探勘技術，將推甄學生資料進行關聯分析，找出學生居住區域、性別、選填志願和就讀意願之間的關聯。接著根據關聯分析結果建立決策樹，以此決策樹預測下一學期個人申請入學可能就讀的人數，模型準確率可達87.5%。

During the effects of the low fertility rate, the Ministry of Education has decided to use the rate of freshman registration as the basis for the survival of private universities. As a result of the increase in the proportion of individual admission, it is the most important issue for universities, especially private universities, to grasp and choose students who are strongly enrolled from the individual admission. In the past, the department chose students through the subjective judgment, but also the lack of follow-up verification. In this paper, by the data mining, the analysis of student data will be analyzed to find out the association rules about students' location, gender, candidate colleges and decision results. Then to build a decision tree based on the association rules, so the decision tree to predict the number of freshmen of individual admission for next semester, the model accuracy rate of up to 87.5%.