應用資料探勘方法探索學系對新生選讀該系關聯規則之研究

A Research of Applying Data Mining Techniques to Explore the Associated Rules to Department Application and Entrance of Fresh Newcomers in Taiwanese Universities

黃冠豪(Guan-Hao Huang)

指導教授 ： 曹以明

義守大學/管理學院/工業管理學系/碩士(2017年)

摘要

近年來，受到少子化的影響，雖然大學新生入學錄取率逐年提升，也因為此有一些學校的一些學系卻因為招收學生人數不足，進而影響到學校經營的困難度。然而，有些學校因為經營上困難，因此必須轉型或者停止招生而影響到師生的權益。倘若能夠了解大學新生對於大學院校和學系選擇的可能因素，並分析兩者之間的關聯或影響程度，當有助於校系經營的發展以及招生策略的擬定。本研究目的為通過任一學校任一學系提供當年或是多年的入學新生申請資料，經過資料探勘的技術來分析出會選擇該學系學生的一些特質，進而找出各階段學生會做各種決定的相關因素是什麼，以此來提供某學系作為招生及資源分配的參考。本研究是利用樞紐分析以及關聯規則的技巧來分析各階段學生的特質，而本研究是以某一個學系某一年度的資料來做試算，經結果發現在第一階段會選擇該校的多以南部的高雄市學生居多，其中又以公立學校畢業的學生占最多數，第二階段數學為後標且總級分為31~35分的學生最有可能會來報名參加甄試，第三階段地區為南部且總級分為31~35以及地區為南部且地區為高雄市的這兩條規則最有可能會參加後續的面試，第四階段地區為南部的且英文成績為後標的學生最有可能成為正取生，而自然成績為後標且總級分為31~35分的學生以及數學為後標且總級分為31~35分的學生這兩個組合則是最有可能成為備取生，而到了最後階段則是地區為南部且縣市為高雄市為其中最有可能就讀該系的組合。綜合上述的結果來看地區為南部的學生以及總級分為31~35分的學生這兩個類型無論是在報名、篩選學生或是最後選擇科系的行為都是最有可能的，而作為組合的規則來看地區為南部且自然為後標的學生在各階段都有很高的支持度，因此可以判斷這類學生有很高的可能性會就讀該科系。

關鍵字

資料探勘 ； 關聯規則 ； 樞紐分析 ； 學系新生入學 ； R程式

並列摘要

In recent years, the colleges and universities in Taiwan have been encountering difficulties in their sustainable operations because of lower sub-replacement fertility and lack of enrollment in some departments though university freshmen acceptance rates have been gradually increasing year by year. When it is difficult for these schools to survive, some have to make a transformation or stop recruiting students. Thus, the rights of both teachers and students are threatened. In case it’s available to map out the possible factors affecting potential university freshmen while they are choosing schools and departments and analyze the correlations or effects between the two, it will be beneficial to schools and departments for their sustainable development as well as forming recruitment strategies. Based on freshmen’s application information for enrollment with a period of one year or several years provided by a department at a university, this study aims to apply the technique of data mining and analyze the characteristics which facilitated the students to choose the surveyed department. Then, we will be able to find out the factors affecting students when they made decisions at different stages. The analytical results of this study can serve as a reference for the surveyed department in future recruitment and distribution of resources.In this study, pivot analysis and association rules were adopted to analyze the characteristics of the students at each stage. The data of a certain department in a certain year was referred to for trial computing. It was found that the students from Kaohsiung accounted for the most to choose the surveyed school at the first stage. Among these students, the majority graduated from public schools. At the second stage, the students with Math reaching 25 percentile and total score level 31~35 were the ones most likely to register for a screening test. When it came to the third stage, the students from South Taiwan reaching a total score level 31~35 and those from Kaohsiung in South Taiwan were most likely to attend follow-up interviews. At the fourth stage, the students from South Taiwan with English reaching 25 percentile were most likely to be on the admission list. The students with Natural Science reaching 25 percentile and total score level 31~35 as well as those with Math reaching 25 percentile and total score level 31~35 were most likely to be on the wait list. At the final stage, the students fulfilling the combination of “South Taiwan” and “Kaohsiung” were most likely to study at the surveyed department. Based on the above mentioned results, the students in the two categories of “South Taiwan” and “score level 31~35” were the most likely to study at the surveyed department either by enrollment, student screening or final department decision. By taking all the factors into consideration, the students from South Taiwan with Natural Science reaching 25 percentile presented a high level of support at each stage. This means it’s very likely for the type of the students to study at this department.

並列關鍵字

data mining ； Association Rule ； Pivot Table ； New students are enrolled ； R Program

參考文獻

張筠昀(2015)，應用藍海策略及資料探勘技術於醫療健康產業及體檢數據庫分析，義守大學資訊管理研究所未出版碩士論文。

楊巧暄(2015)，運用資料探勘與模糊邏輯技術建立糖尿病診斷系統，義守大學資訊管理研究所未出版碩士論文。

王秀槐、黃金俊(2010)，擇其所愛、愛其所擇：從自我決定理論看大學多元入學制度中學生的科系選擇與學習成果，教育科學研究期刊，55(2)，4。

周忠毅(2014)，以資料探勘與模糊邏輯技術建置肝病診斷系統，義守大學資訊管理研究所未出版碩士論文。

徐詠宸(2015)，應用資料探勘與模糊邏輯技術建置投資理財決策系統，義守大學資訊管理研究所未出版碩士論文。