Learning Analytics


Learning Analytics

“ Learning analytics is the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimising learning and the environments in which it occurs. ”

I am interested in how Learning Analytics can motivate learners to engage in learning activities more and/or change how they study. The ultimate goal of my current research is to provide students with effective Learning Analytics visualizations of their activities in learning environment (we use Canvas LMS) to motivate them in changing how they study (i.e. self-regulate their learning tactics) to improve their learning outcomes. We propose to do so through visualizing so called learner model (hence “Open Learner Model”). The Open Learner Models enable students to monitor their progress, and in turn motivate them to improve their learning strategies.

Main idea of the proposed Canvas extensions is to provide scaffolds in the form of OLM to externally-facilitate self-regulated learning by accounting for different cognitive conditions. Learners are generally weak in regulating their learning and require external scaffolds. Students have different goals and we posit that different scaffolds are necessary to motivate individuals with different goals to self-regulate their learning. We can facilitate evaluation decision making by presenting information about learner’s learning progress (in forms of products in relation to standards) through OLMs, such as counts of posts as compared to the class average or to the best individuals within the group, following up on the posts of others, concepts covered in the posts (via concept analysis), etc. If OLM presentation suits individual’s goal orientation it can motivate her to adopt it to monitor and consequently control her learning. The goal is to allow the student to control her learning through adaptation of what they do in discussion groups within Canvas, which can lead to increased quality of learning products (posts) and potential changes to cognitive conditions.

This is the idea, but if and how it works is to be discovered. The truth is that we do not have a good theoretical understanding of all the contributing factors to the model. We aim to build a theory that will inform tool developers how to design effective visualizations. To do so, we build algorithms that extract learning models from log and trace data (including natural language processing, data mining and machine learning), we build prototypes and integrate them into courses, run user studies within existing courses, analyze the data and refine our models. Of course, we share the results with our community through publications.

There are several diferent directions that we are pursuing, for more information check the pages of my graduate students.