網路學習不同於傳統課堂授課,教師與學生之間無實際接觸,因此教師較難掌握學生學習狀況。藉由分析學生線上學習之學習歷程,教師可以觀察學生的學習活動,並對因線上學習活動不積極,而可能落入學習成效不佳的學生,給予鼓勵幫助。 學生線上學習歷程通常為量化之連續數值,因此傳統以統計方法進行分析,但一般統計分析(例如,平均值 、標準差),需具備相當基礎才能理解,對一般教師來說較難有效應用。因此本研究以K-means集群分析演算法對學生各項學習歷程進行集群分析,藉以輔助一般統計方法。集群分析能有效將學習歷程相近之學生歸類為同一群集,且不同群集間具有相異性。 本研究利用統計學之T考驗及獨立性考驗,分析學生在線上學習的活動表現與學生的學習成效之間的關聯性及相關程度。同時我們以資料探勘之關聯規則分析擷取可信度高之關聯規則,提供教師輔導學生之用。 本研究提出之學習歷程分析系統,提供統計及資料探勘技術針對學生的個人背景資訊、作業成績、測驗成績及線上學習紀錄等學習歷程進行探勘分析,提供介面讓教師觀察學生在學習過程中的學習型態變化,並分析其與學習成效的關連性。
Web learning is different from traditional learning. Teacher and students are not face-to-face, so it is difficult for teachers know well with students' learning situation. Base on analyzing students' learning portfolios on learning platform, teachers can observe students' learning activities. Furthermore, teachers can encourage and assist students who might fail the course due to their native learning behaviors. Most of students' learning portfolios on learning platform are continuous values. Traditionally, statistics strategies are adopted to analyse these data. But it’s difficult for teachers who do not have background of statistics to comprehend. In this thesis, K-means clustering analysis algorithm was adopted to analyse students' learning portfolios. Cluster analysis can group students who behave similar learning portfolios in one group efficaciously. In this thesis, t-test and independence test were adopted to analyse correlation between students' learning portfolios and achievements. Also, the association rule was used to discover rules with high confidence. According to these rules, teachers can provide assistance to students with different situation. An integrated learning portfolios analysis system was developed in this thesis. This system provides statistic and data mining techniques to analyse students' personal information, homeworks, test grades, web learning records, etc. Teachers can easily monitor students' learning behaviors in progress, and discover correlation between achievements and these learning behaviors.