HEFCE Project: “Improving Student Engagement and Retention through “human-in-the-loop” Learning Analytics”
The aim of this HEFCE-funded project is to develop novel Learning Analytics techniques, incorporating practitioner knowledge and student insights, to improve outcomes for Undergraduate students.
The significant benefits of learning analytics for improving teaching and supporting learners are well understood (HE Commission, “From Bricks to Clicks”, 2016). The importance of course-specific data, acknowledgement of student learning styles, and appreciation of discipline-specific pedagogical approaches, is critical in developing effective Learning Analytics.
Failure to incorporate practitioner knowledge, and student experience, means the full potential of Learning Analytics is rarely realised, limiting the quality of actionable insights. This presents significant barriers to student attainment, compromising our ability to enact timely interventions for at-risk students.
We propose a “human-in-the-loop” approach, engaging students and course leaders as active partners in Learning Analytics. This enables richer interaction with data, encouraging dialogue around student progress.
One of the primary contributions of our project is a novel software tool, providing advanced interactive data exploration capabilities and providing insight into the analytical processes of course leaders.
Additionally, throughout the course of the project we have been continuously engaging each of our three Faculties, building and testing small prototype tools and reporting/analytics pipelines to suit their varied needs and experience.
The novel software tool described above is an application called Glance. Glance creates “survey dashboards” containing sets of visualisations about student engagement and performance, asking instructors to compare and rank students based on their predicted performance in a given module,
The application provides a framework for:
-
Evaluating the impact of different instructor-facing dashboards, by measuring the accuracy of instructors’ mental models of student performance when presented with each.
-
Capturing instructor predictions about student performance through an intuitive and easy-to-use interface. These may then be measured for accuracy against students’ true performance and used to improve simple machine learning models.
The applications itself has been used for a number of small scale studies here at Newcastle, but was never intended to be Newcastle specific. To that end, it is open sourced under a permissive license, its setup and use thoroughly documented.
We have scheduled several open-call webinars. During the webinar, we will present preliminary results arising from the project, and provide an introduction to the use of Glance at your own institution.
- 9-10am GMT Friday 9th March 2018
- 9-10am GMT Tuesday 13th March 2018
- 9-10am GMT Friday 16th March 2018
To sign up for the webinars, please email [email protected], stating your name, institution and preferred date.
For project enquiries, please contact Matthew Forshaw, Lecturer in Data Science ([email protected]).