Integration of Learning Analytics in Blended Learning Course at a University of Technology

Abstract

The main purpose of this study is to use Learning Analytics to improve the instructional design in an undergraduate Mathematics Education I course. Students enrolled in the course come from varying backgrounds. The Learning Analytics will improve the flexibility of the course and provide a platform to understand misconceptions experience by the students. The integration of computational aspects is necessary to illuminate teaching learning and assessment in a Blended Learning setting. A Blended Learning model is used to teach first year undergraduate mathematics education I course at the School of Education, Durban University of Technology. Students were taught a course in mathematics using a Learning Management System. Data is constructed using items from the discussion forum on Black Board and a post assessment task given to 170 first year Mathematics Education I students. Four levels of Learning Analytics: descriptive, diagnostic, predictive and prescriptive are used to discuss the data set. Activity theory is used as theoretical framework. Mixed methods were used to analyse the quantitative and qualitative data. Student errors from the post-test are categorised as cognitive errors with structural errors, arbitrary errors and executive errors to establish links with pre-knowledge frames and concept representation. The structural errors indicate that the representation of concepts is necessary in the content design of the course. Results show that there are more structural errors than executive and arbitrary errors.