e.g. school registers, class records
Study records, grades, curricula
e.g. ViLLE, Moodle
Login data, submissions, course enrollments, attendance records
e.g. national statistics, open data, questionnaires
Background information, research materials
The goal of learning analytics is to offer tools and methods for monitoring studies and predicting difficulties in learning. Guidance counselling and automating teaching and learning aids are of particular interest in the field. Through these means, we seek tested, justified, and ethical ways to enhance learning, teaching, study guidance, and administrative processes.
A typical approach to creating tools in learning analytics is the development of predictive models from existing data. For example, by examining students' yearly accumulation of study points it is possible to predict the accumulation of upcoming years. This enables early intervention when corrective measures are still useful in preventing problems.
Questionnaires can be powerful tools in gathering data for learning analytics, especially if it can be combined with other data (such as that gained from learning platforms or sensors). Still, questionnaires alone can also yield important results. The attached picture combines two questions picked from a questionnaire measuring study habits of students.
As we can see, students who report having friends they can complete assignments with also report significantly fewer occasions of feeling mentally or physically tired. Therefore, it is clear that formation of friend groups among students should be facilitated.
Continuous assessment and comprehensive collection of course achievements enables the development of a model that can be used to predict course outcomes. The picture shows achievements of the first two weeks of an eight-week course. Each dot represents a student enrolled in the course. The dots are color-coded based on the final grade achieved. What is notable here is that the model can predict 80% of students who are going to fail the course during the first two weeks.
ViLLE automatically recognizes students' learning misconceptions in mathematics based on the information collected from their submissions.
A study showed that the algorithms of ViLLE predict learning misconceptions as effectively as a widely-used pen and paper test. The difference is that automatic analytics enables real time viewing of information without a separate test.
As data gathered from students is at the center of learning analytics it is imperative to take special care of data privacy. We at the Centre of Learning Analytics have always acted according to law and the following ethical principles: