Online Outlier Detection System for Learning Time Data in E-Learning and Its Evaluation

M. Ueno (Japan)

Keywords

elearning, LMS, Data Miming, Learning Histories Data Base

Abstract

Recently, distance education by using e-Learning has become popular in actual educational situations. However, there is a problem that the instruction strategy tends to be one way, and so it sometimes makes the learners bored comparing with usual instruction methods. This paper proposes a method of on-line outlier detection of learners' irregular learning processes by using the learners' response time data for the e-Learning contents. The unique features of this method are as follows: 1.It proposes an outlier detection method by using Bayesian predictive distribution. 2. It is available for small sample, 3.It provides an unified statistical test method of the various statistical test by changing the hyper-parameters, and it provides accurate test results than one of the traditional methods. 4. On line outlier detection is realized on WWW. 5.It assists two ways instruction by using data mining results for the learners' learning processes. 6. The outlier statistics is estimated by considering both the students' abilities and contents' difficulties. This paper evaluated the efficiency of the proposal , and the results show the efficiency of the system.

Important Links:



Go Back