The first international workshop on Machine Learning and Learning Analytics took place last Tuesday 25th March, as part of LAK14. The goal of the workshop was to discuss what benefits machine learning techniques can bring to the field of learning analytics. It was intended as both an introduction to the field for those wishing to start applying machine learning within their own institution, as well as an opportunity for those already applying machine learning in different institutional contexts to discuss important issues.
The workshop was very well attended, with over 50 participants ranging from relative newcomers to experts in the field of machine learning. The day was structured to include introductory talks in important topics of the workshop given by conference co-organisers. This included a talk by Annika Wolff with the theme of understanding student learning behaviour through building predictive models. There were 5 paper presentaitons, 2 of which were given by KMI’s Jakub Kuzilek and Martin Hlosta, focusing on the work of the OUAnalyse project in applying predictive modelling to identifying ‘at risk’ students. Several breakout sessions during the day covered recurring themes of the presentations.
Important issues discussed throughout the day-long workshop included the importance of data within predictive models, how click-stream data such as obtained from a virtual learning environment could best be captured and used for prediction and how to evaluate the effectiveness of models when interventions are made with the goal of changing an outcome predicted by a model. The workshop was co-organised by KMi’s Annika Wolff and Zdenek Zdrahal, along with George Siemens, Carolyn Rose and Dragan Gasevic.
