A growing number of universities are using predictive analytics to accelerate learning and improve student success rates. Find out how HEIs are leveraging algorithms to facilitate early interventions and uncover various pitfalls.
Predictive analytics is used across the higher education sector to predict the probability of future trends that impact student outcomes and success rates. It uses a combination of big data, algorithms and machine learning techniques.
Tech giants such as Netflix and Amazon were early adopters of predictive analytics, harnessing it to track user behaviour in order to sell more products and service.
According to a Zion Market Research report, the global predictive analytics market is on track to reach approximately $10.95 billion by 2022.
How are universities leveraging predictive analytics for student success?
Higher education institutions are using machine learning to understand their students better; both in terms of how they engage with programme content and the university experience as a whole.
By forging a better understanding of what determines success, your institution is equipping itself with the knowledge it needs to build better systems; systems that provide students with more personalised and timely support.
You can use predictive analytics to build a more accurate understanding of the factors that hold students back.
Leading subject matter experts are already talking about how predictive analytics can be used to enhance learning, boost engagement and ensure success.
Find out what they’re saying in this video:
7 ways higher education is using analytics for student success
Although the technology is in its infancy, forward-thinking universities are already reaping the rewards of predictive analytics.
A growing number are using it to do the following:
- Adapt and personalise the learning experience to make it more efficient and effective.
- Identify individuals who are at risk of dropping out by tracking patterns of behaviour, and using these insights to intervene early.
- Accelerate students’ learning by moving quickly through the content they know and providing enhanced support in areas they are yet to master.
- Monitor student engagement and performance in real-time to identity macro trends.
- Gain a better understanding of student comprehension in order to inform assignment, content and programme design.
- Improving the ability to reach the most vulnerable students without hiring more staff.
- Re-allocate time to impactful work such as having transformational conversations with students, and programme mapping.
Some higher education professionals are concerned that monitoring students to such a degree constitutes an invasion of privacy.
Critics also warn that algorithms could in fact reinforce historical inequalities.
So, how can you use predictive analytics responsibly?
Predictive analytics for student success: 3 best practices
Working with predictive analytics can be challenging, so you should always consider these 3 best practices before diving deep into this effort:
Best practice #1: Avoid implicit bias in algorithms
If institutions aren’t careful, algorithms could in fact exacerbate the impact of structural bias. After all, human beings are the ones programming the algorithms in the first place.
It’s argued that if a HEI implements algorithms that use identifiers such as post code, secondary school, or ethnicity, they are at risk of under-serving students’ needs.
Georgia State University (see case study below) intentionally excludes these non-changeable factors from its predictive modelling.
Best practice #2: Give students control over their data
When it comes to gathering data, it’s important to provide students with the option to opt in or out. Ultimately, it’s up to the individual to give consent.
With this in mind, how can you encourage students to opt in?
- Be transparent about how data is used and maintained
- Deliver personalised and tailored content
- Ensure understand exactly what they will be missing out on if they don’t opt in
- Investing in smart, AI-based solutions to automate unsubscribes
Some universities analyse data to find out how students are utilising on-campus or online services. Instead of using personally identifiable information, they monitor class-wide data in order to improve services.
This only applies to the students who have actively ‘opted in’ though.
Best Practice #3: Restrict access to student data
It's unlikely that many staff members will require access to the full breadth of student data. Institutions need to have data governance practices in place to ensure that the right people see the right information.
For example, one employee may only require access to attendance statistics whereas another may also need to see metrics on how individuals are using the library service.
Ultimately, predictive analytics can help HEIs, like the one below, move from an institutional mindset to a student-centric one.
Predictive analytics case study – Georgia State University
Georgia State University (GSU) started using predictive and data analytics for student success in 2012. Since then, it’s had over 250,000 one-to-one interventions with students as a result of system prompts.
GSU is tracking 800 different students every day and predictive analytics is enabling the advisory team to deliver advice to students at scale.
One way GSU uses predictive analytics is to identify students who register for courses that don’t apply to their degree programmes.
Before the use of predictive analytics, some students would opt for the wrong course without anyone noticing. They would have to leave the course after starting the wrong one and enrol on an appropriate option.
Now, an adviser receives an alert when a student signs up for the wrong course, enabling them to promptly set them on the right track.
Over 2,000 corrections were made in 2021 alone.
As well as alleviating specific issues, GSU has experienced a university-wide improvement. Its four-year graduation rates have improved by seven percentage points since predictive analytics was implemented and students are graduating quicker.
“ Now, we have begun to deliver the guidance students need in a timely fashion, and we’re doing it across the board for every student. It’s having a big impact, especially for low-income and first generation students.” - GSU
If you’re looking to harness the power of predictive analytics, or would like some guidance on how to digitise your admissions and student management processes, check out our webinar: "The future of higher education admission."
Frequently asked questions
How do you develop predictive analytics?
When implementing a predictive analytics strategy, you must first define what you want to achieve (e.g. higher retention in X programme). Then, collect relevant data and improve the quality of the data by testing it.
How can big data help higher education?
Big data can help higher education organisations to improve their systems, processes and programmes by equipping them with the insights required to identify risk and success factors.
How do I choose a good predictive model?
When opting for a predictive model, consider factors such as computation and performance, as well as the quality of the bias variance threshold.
What are the three pillars of data analytics?
The three overarching pillars of data analytics are agility, performance and speed.