依據教育部資料顯示，本國大專校院數量不斷增加，大學日間部學生招生人數亦隨之增加，然而，根據內政部統計資料表示，本國出生人口總數相對持續減少中，以致大專院校招生情況日益嚴峻，亦影響自國小、國中至高中職未來招生困難之處。對上述問題，本研究運用資料採礦技術，以某大學94-96學年度大學日間部學生資料為例，分析學生歷史學籍資料，以建立學生流失預測模型並進行預測評估，同時探索流失學生之表徵。本研究藉由測試驗證預測模型效益之可行性，經研究顯示，成績為學生流失的主要影響因素。最後，本研究針對研究結果，提出相關建議，以供學校參考以改善學生流失之情況。 According to the data from Ministry of Education, the number of college increase constantly. The enrollment of students in the day school also increases. However, the statistics from Ministry of Internal Affair showed that total birth rate of Taiwan is decreasing. Therefore, the enrollment status is getting tougher. This research is based on the day school student enrollment data between 2005～2007 in a University. By using the data mining techniques to analyze the historical records of students, this research tried to find the potential reasons of why students drop out, and to create a prediction model for student drop out. The analyzing and mining results from this research will provide relevant suggestions to help the university to improve the situation of student drop out.
