Statistical analysis for building a personalized online learning system

Abstract

Use of computers and new technologies have become a crucial part of teaching and learning at all levels. Online courses are widely available and the number is growing exponentially. One advantage of most online learning courses is that students’ performance can be made tangible and measurable. We analyze databases from two e-learning platforms developed for two independent courses at The University of Hong Kong, namely SCNC1111 (Scientific Method and Reasoning offered by Faculty of Science) and NURS1516 (Introduction to Statistics offered by School of Nursing), that record traces of students’ online activities on the platform. Item Response Theory models are applied to the two databases to estimate students’ ability based on their performance on solving questions online. It is found that active participants showed improvement in the mathematical capacity in both courses. Moreover, significant differences in students’ online behaviour are detected probably due to the differences in the settings of the two courses. The results of the analyses based on the two datasets will be summarized and compared in this thesis. The direction to the development of a personalized e-learning system will be discussed based on the results of the analysis.