MOOC Design and Learners Engagement Analysis: A Learning Analytics Approach

With the low learner completion rate being one of MOOCs' major criticisms, several research have found that learners' engagement is a key success factor in MOOCs. This paper proposes a learning analytics process to discover interesting patterns found in MOOCs and their learners data using two important dimensions: i) course structure & design, and ii) learners' performance & engagement. The relationships of both dimensions are also investigated. More specifically, machine learning and visualization techniques have been employed to discover and analyze these aspects with the datasets extracted from Thai MOOC National Platform. Regarding the course structure & design dimension, it has been found that the total course length and the required learning effort can be used to cluster Thai MOOCs into short, medium-length and long course clusters. In addition, by analyzing the defined learning outcomes, classified by means of Bloom's taxonomy of cognitive process, we have clustered Thai MOOCs into understanding-focus, applyingfocus and multiple-skill-focus course clusters. Considering the learners' performance & engagement dimension, grades and attempt rates have been used to cluster Thai MOOC learners into sampling, targeting and comprehensive learners. Statistical tests have also confirmed the significant relationship between course structure & design and learners performance & engagement. Hence, valuable insights and recommendations have been drawn to better design more engaging MOOCs and to better support learners.