Machine learning, used within Learning Analytics, gives us new insights into education processes. Modern AI can predict student behavior and by doing so improve the entire education landscape.

In November, with a successful review meeting, we finished our latest Horizon 2020 research and innovation project ENVISAGE that kept our attention for the last two years. The idea behind the project was to use machine learning, developed for digital games, to make the educational processes in schools and universities more efficient for students and teachers.

Why is Game Analytics a good choice to bring machine learning into education?

The education landscape is changing. New technologies are transforming the way how we study and how we will learn in the future.

Game Analytics is already highly developed when it comes to profiling users, predicting their behavior and adapting the games personally to each individual player. Applied in education, Game Analytics can be used to analyze virtual labs for studying, including remote studying. The technology also enables detailed tracking of student‘s behaviors to adapt learning content to every student‘s individual requirements.

What we did in the ENVISAGE project to adapt AI for education

In this project, we used our long-term experience from the gaming industry to support the educational processes more efficiently.

Our main tasks in the ENVISAGE project included:

  • Adapting machine learning algorithms for the education sector
  • Collecting learners‘ anonymous data with help of our server infrastructure and the accompanying SDKs
  • Developing new machine learning algorithms to make exact predictions about students’ behavior
  • Predicting at-risk behavior of students as well as positive influences on students. This way, we can help them to become more motivated and enjoy learning
  • Establishing business cases, analyzing the market and competitors, and developing new ways to use the technology in our products and for other industries Infrastructure for Learning Analytics and At-Risk Student Prediction. Infrastructure for Learning Analytics and At-Risk Student Prediction

Example Use Case: At-risk Student Prediction

Virtual labs and learning software produce dozens of events per student per lecture which describe their learning behavior and progress. This data can be tracked in form of events via our SDKs and tracking infrastructure as shown in the figure above (Step 1). Typical events are the beginning or end of a task, answering questions in an exercise, or interactions with the user interface. These events are then aggregated and transformed to construct datasets for machine learning (Step 2). To improve learning, in particular in remote settings, it is helpful to predict a student’s engagement. I.e., is the student more or less likely to return to a learning app. This is also referred to as “At-risk Student Prediction”. Here, a machine learning model can be trained that predicts a student’s at-risk likelihood. Once this prediction is available for each student, different actions can be performed. For example, if a student is likely to leave because the difficulty is too high, we can use dynamic difficulty adjustment to adapt the content. The same works the other way around if the learning material is not sufficiently challenging. In other settings, we can inform the educator about the at-risk behavior of students, so that more attention can be paid to this group of students.

Of course, AI in education should not and will not replace a teacher in any time soon. We are convinced that modern tools, based on AI and machine learning, will support teachers in the classroom. With this technology, teachers can meet the requirements of every student individually and help every one of them in the best possible way.

Our success in this project with Learning Analytics

In the project, we did not only find many ways to use AI in education, but there was even more we did. We have started to organize workshops with customers on „AI in Education“ where we developed different use cases for each customer. We also integrated new features into our product and validated these in the gaming area.

Validation in digital games was very important and beneficial, as games offer much bigger data volumes compared to virtual labs in schools currently. This way we showed that our solutions are working at a large scale and can be used safely and reliably in educational facilities.

You can find more information about case studies in games in our blog under:

Awards and benefits from the ENVISAGE project

The ENVISAGE project did not only help us to explore a new market but we also received very positive feedback from advisors, reviewers, and representatives of the Europen Commission. On top of that, the project got us nominated for the Innovation Radar Prize 2018 in the category „Best Young SME“. All of us are very proud of this success and we look forward to the new opportunities within the Horizon 2020 program. On behalf of the team, we would like to thank all of our consortium partners again for the high-quality work and great collaboration.

Members of the ENVISAGE consortium in Luxembourg after the successful review meeting.

To learn more what offers for the education sector, please visit

or reach out to us directly. And don’t forget to subscribe to our newsletter and follow us on Twitter, Facebook, LinkedIn, or XING.