以修課資料為導向利用分群化發掘學生適性通識課程

Course Taking Data Oriented to Find Students’Adaptive General Education Courses through Clustering

本研究以學生的修課資料為探勘資料來源，每一筆修課資料記錄學生曾經修讀過的課程項目與等第成績，以k位學生為探勘目標，k≧1，文中以課程等第相似度做為分群的準則，設計一個以學生為中心分群化修課資料成k個群組的方法，從各群組中找出中心學生與通識課程項目之間關聯性，分別做為判斷k位學生適性選修通識課程的依據。文中將k位學生的修課資料，分別設定為一群組的中心點，計算修課資料與各群組中心點之間的課程等第相似度，並將修課資料歸屬於課程等第相似度最大的群組中。每次分群化之後計算整體課程等第相似度的總和，若目前分群化的整體課程等第相似度總和大於之前的分群，則將目前分群的中心點取代之前的中心點。在分群化之後的群組中計算未包含於探勘目標中心學生之修課資料的其他通識課程的出現比率值，找出比率值最大的前j項通識課程，j≧1，稱之為學生適性通識課程，或是針對學生目前學期可選修的通識課程進行出現比率值的計算並進行推薦。文中根據提出的方法，設計與建置一個發掘學生適性通識課程探勘系統，本研究探勘結果，對學校課程管理單位推薦學生適性通識課程項目，將可以提供相當有用的參考資訊。

This paper uses course taking data as the source of mining, and each course taking data records the course items and grades that students have taken. Taking k students as the mining target, k≧1, the similarity of course grades is used as the criterion for grouping, and to design a student-centered method to cluster course taking data into k groups. We find relevance between central students and general course items from each group, and as the basis for judging the suitability of k students to take general education courses. The course taking data of k students are set as the center points of a group, and to calculate the similarity of course grades between course taking data and the center point of each group. Then assign the course taking data to the group with the greatest similarity in course grade. After each clustering, the sum of the overall course grade similarities is calculated. If the sum of the overall course grade similarities in the current clustering is greater than the previous clustering, the center point of the current clustering will replace the previous center point. We calculate the occurrence ratio value of other general courses that are not included in the course taking data of students in the mining target center in the group after clustering, and to find the top j general education courses with the largest ratio value, j≧1, which are called students’adaptive general education courses, or it calculates the occurrence ratio value and make recommendations for the general education courses that students can take in the current semester. According to the proposed methods, a mining system for finding students’ adaptive general education courses is designed and built. The results of mining can provide useful reference information for school curriculum unit to plan students’personal adaptive general education courses.