Machine learning for higher education

Students at risk

The financial impact on students who drop out of university and the flow on effect for education providers is significant.

Over a four week period, a team of Jade’s data scientists built, trained, and optimised a machine learning algorithm to identify students at risk of dropping out.

Sounds simple, right? Well, here’s how they did it:

A four-step guide to predictive AI

1. Extracting and cleaning data:

This involved finding, consolidating, and cleansing the relevant student data. This produced over 1,000 tables of student data to analyse.

2. Visualising and analysing:

With usable data available, visualisation helped distinguish which factors were likely to impact a student’s chances of dropping out in one way or another. Influencing factors weigh in differently for each student and interpreting them accurately in each situation is a key difference between machine learning and more traditional analytics.

3. Creating a model:

In this step, we applied various algorithms to the data then tested and refined them in multiple iterations until we were satisfied the model explained the predictions reliably and accurately. Learn more about how these algorithms work in part 1 of our series, Thoughtful machines: What is artificial intelligence?

4. Integration:

The last step set up live data feeds into the model and allowed it to make predictions on a student's likelihood of dropping out.

The output:

Using this method, a percentage score indicates a student’s chances of dropping out. This could then be passed back to the university. From here a university can assess the dropout risk for each individual and offer students appropriate support to help them stick with their studies.

A wider scope

Machine learning isn’t just limited to education - it has application in other industries too. You could, for example, use similar indicators to predict the chances of a customer unsubscribing from a service in an effort to reduce customer churn rates.

So, how do you get started with AI?

In the next edition, Getting started with AI, we’ll take a look at two way to get the ball rolling with limited effort and investment.

See you then.

Got A Burning AI Question? Ask A Jade Expert!

Jade Software

Jade brings new ideas to life in industries including energy, insurance, agritech, and retail. With a UX-led approach to defining problems and a methodical, iterative process for solving them, we change people’s real-world experiences with whatever technology fits best.

Thousands of companies around the world rely on Jade every day of the year.

We are headquartered in Christchurch and have offices in Auckland, Dunedin, Sydney, Melbourne, and the United Kingdom.