Increasing digitisation of student records means predictive analytics is expected to transform teaching and become a key tool in learning more about students.
Subscribe now for unlimited access.
$0/
(min cost $0)
or signup to continue reading
Predictive analytics is a process in which data collected about the student – typically attendance, subjects taken, assessment – is used to understand learning patterns, identify skill gaps, predict performance and identify learning opportunities.
Teachers and administrators have always analysed student data, but as more student records become digitised – for example, using a consistent online marking system school-wide – there are more opportunities for analysis. Digitised data is also quicker and easier to analyse.
Predictive analytics software assesses student data for strengths and weaknesses, and predicts whether a student is likely to pass or fail a subject, and can even suggest a change in teaching strategies. It can also compare a student's performance against that of their classmates.
Peter Kokkinos, general manager of education software provider Desire2Learn, says analytics reveal trends and allow teachers to make informed decisions about what to teach and how to teach it.
‘‘Is the class trending to pass or fail? Why? One thousand students found a particular topic in a course difficult, and became disengaged. Why?’’
But surely teachers themselves notice when and why students are disengaged?
‘‘In higher education, some lecturers may have thousands of students, and remote students as well, so it’s impossible to keep track of them all [personally],’’ Mr Kokkinos says.
The idea is to learn things about every student that can improve their education, says Professor Simon Kaplan, director of Skills and Industry Transformation at NICTA, Australia’s information, communications and technology research centre.
‘‘Predictive analytics gives richer educational opportunities to more students, guiding them on different paths depending on their capabilities and skills. But predictive analytics is still in its infancy in Australia. The average university only collects sparse online data on its students.’’
Quality data
The quality of any analysis depends on the quality and quantity of data used. The more data collected, the more accurate a picture can be drawn and more accurate predictions made.
Useful data on students, other than assessment and attendance, includes how often and when they read university emails, how often and when they log on to the university’s system, when they submit assignments, and when they are on campus, Professor Kaplan says.
Most universities have this type of data available today, but it is often in isolated systems and difficult to analyse, says Glenn Wightwick, chief technologist and director of IBM Research Australia.
‘‘There is enormous value to be gained from the analysis of data that is typically locked up across multiple systems in their organisation,’’ he says.
The first step is to articulate what the big challenging problems are, identify relevant data, and develop an analytics strategy to gain insight to the problems, he says. ‘‘The idea here is to quickly create some insight that wasn’t necessarily obvious to the organisation.’’
Dropping out
Predictive analytics can improve the chances of a student completing their course, by identifying students at risk of dropping out and identifying potential interventions such as tweaking the curriculum or timetable.
‘‘Dropping out is a big issue in TAFEs or private colleges,’’ says Nick Stanley, CEO of Sky software, an Australian education software company based in Geelong, Victoria. ‘‘There are funding models where an institution won’t get funding if, for example, a certain amount of students drop out. Traditionally you see that after-the-fact, but with predictive analytics you can see a pattern that may alert you,’’ he says.
‘‘Attendance, assessment, lateness, non-payment of fees - by themselves not a big deal, but join them together and you can get an idea of the students that are likely to drop out.’’
If a student is at risk of dropping out, then a teacher or administrator can help, Mr Stanley says. ‘‘If we’re funding this student, we want the student to achieve. Predictive analytics can answer: Did we deliver [the right] education? How many competencies did they achieve? Did the student graduate? How many completed their studies?’’
IBM used analytics to help Edith Cowan University in Western Australia improve its student retention and support programs.
Since 2008, Edith Cowan University has been working to consolidate data across multiple faculties, schools and support centres, Mr Wightwick says.
‘‘The analytics program is enabling the university to gain far more insight into their students, how they are learning, whether they are at risk of dropping out or showing signs of requiring extra support. From this information, the Student Services team can be mobilised to engage with and help the students.’’
Course content
Predictive analytics systems can also track how students interact with course content, says Kaplan. ‘‘This is easy with MOOCs (massive open online courses). Online interactions mean every student’s access to content can be tracked.’’
Some analytics software allows teachers to create their own online content from MOOCs and websites. Teachers can track which online class activities or school resources students are accessing, which tasks are completed, how long they are spending on a topic, which videos they are watching, how much time they spend on homework, and even if they are wasting their time on Facebook, unless of course it is a teacher-approved Facebook page.
‘‘You can think of each student as a data stream from years 1 to 12. Students who do well have a different study discipline, and their behaviour patterns can be examined,’’ says Kaplan.
Teachers can also see how students are collaborating, says Kokkinos.
‘‘Traditionally, content goes to a learning management system [administrative tracking system], but we are now moving to the idea of an integrated platform where students share learning and collaborate, create a learning portfolio and build on it.’’
But the ease of capturing student information and storing it online brings privacy concerns with respect to accuracy, confidentiality and access, particularly in the light of new privacy legislation in March.
CASE STUDY
Brighton Grammar boys’ school in Melbourne uses the Desire2Learn platform to provide students, teachers and families with data about student performance.
The school has completely overhauled its online learning environment, with student assessments and course content all online.
‘‘It’s boosting a teacher’s ability to identify students who require extra guidance,’’ says Cameron Taylor, e-learning co-ordinator.
‘‘Teachers can see what resources students are accessing, and how they are tracking through the course content. The vice-principal and other senior staff use the platform to monitor anomalies for potential class or teacher issues.’’
Teachers already keep track of students without analytics, but sometimes their ‘‘gut feeling’’ means issues can be overlooked, says Taylor. ‘‘Using predictive analytics is more efficient because no student goes under the radar.’’
The school wants to get a deeper understanding of exactly where each student’s weaknesses are and then identify and allocate specific resources and tools for each student, aimed at addressing the problem areas.
‘‘Predictive analytics will help us identify any student issues earlier, before assessment takes place. It can put up a red flag,’’ says Taylor. ‘‘We will get a whole snapshot of the student. Previously, each teacher could only see the results for their own subject. Now they will have an understanding of the whole student. There could be a trend showing through all subjects, where previously the teacher would not be aware of this. We can predict if the student is suited to a particular course, and then suggest suitable university courses. We can predict if the student will do well in physics or chemistry, or is suited to any of our online learning programs.’’
LINKS
