Using AI and machine learning in education: predicting student outcomes


Improving student outcomes is at the heart of most educational institutions – and the ‘lifeblood’ or ‘data’ that flows through the institution, is now providing an intelligent way for progressive educators to pinpoint gaps in learning and identify at-risk students that would otherwise get left behind. Read on to find out how.

Today, the student data sets that educational institutions have at their fingertips are bigger than ever before, and growing by the day. Lecture clock-ins, VLE logins, library passes and digital access to course material etc., are all captured to help staff monitor and track vital touch points and measure student engagement.

Whereas many educators were once at risk of ‘drowning in data’, today’s student information systems allow users to efficiently manage a wealth of relevant data for each and every student, using dashboards within the software to reduce reporting times and share student insights with all relevant parties.

Nescot College, however, is currently piloting a project to use its data in a far more powerful way to generate student insights. Using predictive analytics software to retrieve and analyse data from a variety of sources, including its student information system, Nescot is exploring how to predict learner outcomes.

Predicting student behaviour

Nescot starts to predict student behaviour

As part of the project, Nescot is looking at student withdrawal data to see whether they have made it beyond six weeks, and examining any characteristics of the students that did not.

Learners don’t withdraw because of one characteristic; it’s typically a combination of results, which is why the probability trees within the software are the ideal visualisations to allow the Nescot team to drill down into different characteristics to identify areas for additional support or intervention in order to improve drop-out rates and learner outcomes.

It’s possible that sharing these analyses with students in future year groups may help them assess the advantages or disadvantages of their study and lifestyle choices, based on the historical data from students on their course.

In the meantime, the team at Nescot is also now analysing factors such as levels of attendance, deprivation, parental engagement, gender, and ethnicity to feed their predictive analytics and generate insights that will help them improve outcomes for all student groups.

Using predictive analytics to identify and support at-risk students

In partnership with Panintelligence, Tribal is developing ways to help more educational institutions like Nescot tap into the intelligence and insights available to them with increasingly sophisticated predictive models.

Using machine learning to study the past and make predictions about the future, educators will be able to get a greater understanding of student behaviour, much more quickly and in time to make a difference by:

  • Identifying at-risk students as early as possible to make interventions to stop them dropping out.
  • Predicting which students have the best chance of passing their exams and which might need extra support.
  • Pinpointing gaps in individual students' learning and adapting teaching style and content to improve their learning experience. 

Panintelligence has created an easy to use tool for staff to understand and challenge the data they are examining. Making it easy to understand, trust, and apply the intelligence from the reporting dashboards, the software empowers users to predict future outcomes with increasing certainty, and tailor their teaching and interventions accordingly.

How predictive models could be used in the future

Operational teams at the college are also now using predictive analytics, alongside their own knowledge and expertise, to challenge business decisions. In the not too distant future, cobotics (the collaboration between humans and robots/machine learning) could also help operational teams develop recruitment strategies.

And as predictive analytics software becomes more prevalent in the education sector, high school students could even use predictive modelling insights to choose the best-suited courses to their interests and abilities.

Here at Tribal, we truly believe that with time and financial investment, predictive models could transform the education sector by improving individual learning experiences so that increasing numbers of students achieve their best possible learning outcomes.

To find out more watch this workshop recording, where Ken Miller from Panintelligence explains exactly how they’re analysing Nescot’s data to predict student outcomes.