Diagnosis of students’ online learning portfolios

Abstract

Online learning is different from the instruction provided in traditional classes in that the teachers and students do not have actual contact. Thus, the teachers can have little control over the students' learning situations. However, the teachers can observe the students' learning activities by analyzing their online learning portfolios. These portfolios usually comprise quantified serial of information figures. This study used data mining techniques such as a K-means cluster analysis algorithm in order to manage a cluster analysis of the students' various portfolios and to support general statistical methods. The cluster analysis effectively categorized students with similar portfolios into the same cluster; additionally, different clusters revealed different characteristics. In addition, this study also used a t-test and a test of independence of statistics in order to analyze the correlation and degree of correlation between the students' performance in online learning and their learning effects. Further, this study used statistics and a data mining technique to obtain a mining analysis of the portfolios, namely, the students' assignment scores, exam scores, and online learning records; it also applied a proposed interface to enable the teachers to observe the change in the students' learning situation during the learning process in order to analyze its connection with learning effects.