近年來台灣的大學生延畢人數有日漸上升的趨勢，根據教育部統計，在民國104年時達到4.5萬人。一般大學生常因為社團、打工而忘記自己學分，導師方面追蹤學生修課狀況不易，目前學校機制利用期中期末考後有1/2修課學分被當而預警。本論文利用淡江大學資訊工程學系歷史學生資料套入風險管理延畢預警機制驗證預警準確度，發現有延畢生在大學四年內沒有預警的情況，和非延畢生卻被誤判為延畢生而預警比率高的問題。 本研究先探討延畢生在大一及大二的表現狀況探討延畢生沒有被預警的情形；在大三及大四部分，將探討是否因為預警條件不夠嚴謹而導致延畢生沒有被預警原因。再利用前述修改的預警條件，將所有學生套用預警系統，接著利用逐步迴歸法找出延畢生與誤判學生差異的關鍵因子，根據找到的關鍵因子制定白名單排除條件，將誤抓學生解除預警的動作。 In recent years, the number of students in Taiwan university who delayed graduation is an upward trend. The school is used the mechanism exam or the final exam doesn’t gain 1/2 of credits to warn. In this research, using of student data who was in Department of Information Engineering of Tamkang University to the warning system which has made by Wei-Chong Zheng verify the accuracy of the warning. There was found that there was not warned the student who delayed graduation in the university, and the students who was graduated on time is high ratio of falsely early warning. In this research, we explore whether the students who delayed graduation is good academic performance to is not warning in the first-year or second-year. And we explore whether the warning rule is not rigorous to does not be warned for the student who delayed graduation in the third-year and fourth-year. We used the warning rule to verify the warning system warned performance for all of the students. First, we use the Stepwise Regression to find the factor that between the student who was graduated on time and the student who delayed graduation. Second, find the best grades by the student who delayed graduation to set the whitelisting rules. Last, if some students reach the whitelisting rules, we will remove the warning. The warning system make more accurate finds the student who delayed graduation.
