以資料探勘技術分析學習評量資料-以國中力與運動概念為例

Data Mining Technique Analyzes the Data of Learning Performances－Taking the Concept of Force and Motion in Junior High School as an Example

王富民

指導教授 ： 謝建成博士

國立臺灣師範大學/教育學院/圖書資訊學研究所在職進修碩士班/碩士(2009年)

摘要

本研究旨在以資料探勘（datamining）技術,分析不同學習成就學生學習評量資料，利用關聯法則之演算法，建構相關學習概念圖。
並設計學習障礙診斷測驗，從學生測驗的答題資料，診斷學習者的學習障礙與迷思，結合建構之概念圖與學生的迷失概念，進而建立「補救學習路徑」，並再利用統計方法來分析，學習者在透過「補救學習路徑」引導後，是否可以顯著提昇學習者的學習成效。
本研究資料探勘所採用的樣本，為某教科書發行機構辦理之96學年度全國性模擬測驗資料，扣除缺考人數考生共30132位。學習障礙鑑定及「補救學習路徑」引導後學習成效評估對象為台中縣某國中三年級已完成國中「力與運動」相關概念學習兩個班級的學生，每班30人共60人。本研究探討的相關概念以選用之模擬測驗所涵蓋之「力與運動」相關概念為主。經過相關的分析後，我們可以得到幾點結論：
一、利用資料探勘技術關聯法則之相關演算法，分析學生學習評量資料可以協助快速建構出中、低學習成就學生相關之學習概念圖。
二、學生學習評量資料建構之概念圖與一般教科書所安排的學習順序有很大的不同，一般教科書安排的學習順序從基本概念開始再到應用概念，學生建構之概念圖發現學生適合先學應用概念再學較難理解的基本概念。
三、中、低學習成就學生經「補救學習路徑」引導後，學習成效有顯著進步。
四、中分群學生經「補救學習路徑」引導後，學習成效最佳。
五、本研究之方法可提供教師及測驗機構參考，將相關評量資料分析後提供學生適性化的建議。
六、本研究相關演算法可提供數位學習系統開法者參考，建構適性化之補救學習系統。

關鍵字

資料探勘 ； 學習概念圖 ； 迷失概念 ； 補救學習 ； 適性化學習 ； 力與運動

並列摘要

By means of data mining technique, this research aims to analyze the data of individual students’ learning performances, trying to construct the concept map in terms of algorithms of association rule and designing a diagnostic assessment of students with learning disabilities. We can diagnose students with learning disabilities and myth from their answers in questions of the assessment. Remedial Instruction Decision Path is constructed by the combination of the concept map and student’s myth concept. After students are instructed by Remedial Instruction Decision Path, it is statistically analyzed whether students’ learning performances are enhanced.
This research samples 30132 students’ answers of questions in a certain national simulated test. The objects of this research are third-grade junior high school students in Taichung county who have been taught the concept of force and motion. The concept of force and motion in the questions of the simulated test is adopted to analyze in this research. After the analysis, we are capable of having the conclusion as follows:
1.The algorithms of association rule in the data mining technique are able to analyze the data in the assessment of individual students’ learning performances and to construct the learning concept map of middle and low school achievers.
2.The concept map constructed by the assessment of individual students’ learning performances is quite different from the learning order arranged in textbooks. The basic concepts are taught before the applied ones in the latter while the basic concepts are taught after the applied ones in the former.
3.After the middle and low school achievers are instructed by Remedial Instruction Decision Path, their learning performances are improved obviously.
4.The learning effect of the middle school achievers instructed by the Remedial Instruction Decision Path is the greatest.
5.This research approach can be used by teachers or institutes for tests for reference. The analyzed data in the assessment of individual students’ learning performances can provide students with adaptive suggestions as well.
6.The algorithms adopted in this research can be offered to the developers of digital learning systems so as to construct an adaptive remedial instruction system.

並列關鍵字

data minig ； concept map ； misconception ； Remedial-Instruction ； personalised learning ； the concept of force and motion

參考文獻

蔡昆諭（94）。國中學生力與運動的迷思概念。國立臺灣師範大學物理學系在職進修碩士論文。

陳嫻壎(2005)。學習概念相關性之認知結構評量研究-以資料聚類

朱瑀馨（2007）。淺談運用資料探勘技術於財產保險業顧客關係管理之效益。淡江大學保險經營碩士班碩士論文。

黃福坤（2003）。國立台灣師範大學物理系 物理教學示範實驗教室網站。http://www.phy.ntnu.edu.tw/demolab/phpBB/viewtopic.php?topic=7306

高慧蓮、許茂聰、蘇明洲(1993)。一個六年級學童呼吸作用概念之個案研究。屏東師院學報，18，533-570。