Today, successful digital enterprises in every industry require a digital strategy that recognises the value of data, and digital institutions now need to embrace big data too. Doing so will enable the institution to take advantage of emerging techniques and approaches such as Gartner's Insight Engines to generate actionable insights, improving equality and efficiency.
Data strategies need to take account of how data is collected, processed, stored, and acted upon, as well as its governance, provenance, and quality concerns, and the data literacy of both staff and students.
It is no longer enough for any organisation to rely solely upon data specialists working in silos. An understanding of the value of data to all institutional endeavours must be pervasive, just as wider digital literacies and IT skills have been argued for across the education landscape.
Data is everywhere. Data about learners, researchers, lecturers and resources, about their interactions with each other, with institutional systems and with all the digitally-enabled services across the institution.
Of course, it is not sufficient simply to focus on exposing, collecting, storing, and sharing data in the raw. It is what you do with it (and when) that counts – for example, deploying analytics platforms (including some which 'learn', thanks to advances in machine-based learning) that derive actionable insight, and configuring systems, services and processes to respond in an agile fashion (to become better optimised and personalised).
Learning analytics - the application of analytics in an educational context - is beginning to show real signs of success in areas such as student retention. Read our companion report reviewing the state of UK and international practice in learning analytics to find out more.
In the next few years it will start to drive the delivery of new adaptive, personalised learning experiences based on user behaviour and to recommend (or even instigate) interventions for learners at risk of non-completion, based on early warning signs of disengagement.
Data from internet-connected devices and infrastructure has the potential to revolutionise the management of an institution’s physical estate - just as it has in any other organisation.
As a side-effect of increased data literacy, some commercial organisations are starting to expand into the data economy (even if data is not their core business) by spinning out new companies to profit from various stages along the analytics continuum – selling the insight, the data, and/or the algorithm into which anyone’s data can be fed.
The first two of these are unlikely ventures for an educational institution but the intellectual property held within analytic algorithms can become an asset that is well worth exploiting (especially where the research and knowledge tied up in it has the potential to accelerate the time-to-value for consumer analysts).
“This report usefully and succinctly highlights the challenges and opportunities that the increasing range of data available about education provides. Over recent years, data that were once collected primarily for funding and regulatory purpose have been increasingly shown to offer enormous potential to support good decision making, in conjunction with other emerging sources of information.
Right now, in the face of unprecedented HE sector change, we must all work together to take the next step – to fully leverage the true full value that business intelligence techniques and tools can bring when applied to the vast array of data we can now access.
Developments like the HEDIIP Data Language project, Heidi Plus and Heidi Lab1 show the power of working together as a community. Business intelligence is just the beginning of a journey, and now it's very clear that we're on that journey together.”
Jonathan Waller, director of information and analysis, Higher Education Statistics Agency (HESA)
As more services become digital and more devices become internet-connected, more data is generated. This data gives tremendous opportunities to improve decision-making, either through insights into historical data, or through the use of machine learning techniques to generate predictive models.
However, once data reaches a certain size, in terms of one or more of the big data Vs, traditional approaches such a relational databases and desktop analysis tools no longer suffice. The most widely accepted four big data Vs are defined as follows:
- Volume: the scale of data
- Variety: different forms of data
- Velocity: the speed at which data is streaming
- Veracity: the uncertainty of the data
NB: Some authorities would also add other Vs - including value and variability.
Once the size of data, measured by volume, variety and velocity, becomes too complex to manage with traditional data tools it becomes big data.
Traditionally, data used for decision-making has been highly structured, and usually relational. Big data sources, however, are often semi-structured or unstructured, and stored in document stores rather than traditional databases.
Important technologies in this space include the Hadoop framework for managing distributed data sets, NoSQL3 databases, techniques such as map-reduce4 for parallel processing big data sets and concepts such as the data lake5 - a large unstructured data store.
Key areas of impact include:
- Business intelligence - using data from an ever wider range of sources to make business decisions
- Predictive analytics to enhance retention and academic success
- Analytics for pathway planning
- Adaptive learning
In research and knowledge-sharing data itself becomes the currency.
Most institutions are already using business intelligence tools to help them analyse some of their data, often backed by a traditional data warehouse.
Many are already looking at how they can exploit data about learners and their learning to provide better educational experiences, leading to enhanced retention and achievement.
Big data storage
To exploit big data for data-driven decision-making, institutions need to begin to move beyond transitional data warehouses, and begin to consider how to deploy big data storage management solutions.
Actions include developing strategies to collect data from all relevant sources into a data lake, and developing approaches to transforming the data lake into a meaningful structure for accurate and real-time decision-making.
Part of the challenge will be combining data from disparate sources in a meaningful way. Where possible, the use of common unique identifiers simplifies this although, in the big data world, it is understood that this is often not going to be possible. Software suppliers such as Talend and Trillium are increasingly providing solutions to mitigate this problem.
Institutions should consider how accurate data must be for them to make meaningful decisions - with big data it rarely needs to be 100%.
Data generally comes from a variety of sources and is owned by various stakeholders, while analytics and interventions based on the analytics can also affect many different people across the institution.
Cross-institutional learning analytics programmes sponsored by a senior executive are therefore likely to be necessary to ensure that the data sources and the necessary pieces of the technical architecture are in place, together with the various policies and processes required.
Human intervention on the basis of predictive analytics is likely to mean changes to the roles of teaching and support staff, with appropriate training put in place so that they can interpret the dashboards accurately and intervene with students effectively.
Predictive learning analytics
One of the emerging applications for big data in education is the use of predictive learning analytics. This uses data about students and their learning, and can be used to identify at-risk students so that interventions can be taken to reduce attrition rates.
The student engagement analytics diagram below shows some of the data and touch points that could be used to determine a student’s level of engagement (taken from the student engagement analytics section in our guide to relationship management).
How this affects you now
IT services need to start planning their big data infrastructure. They will need to understand their data sources and ensure that staff have appropriate expertise in core technologies and techniques such as Hadoop and NoSQL.
They should also begin planning their big data warehouse approach, including how data lakes are going to be transformed into data that can be interrogated to allow actionable insights.
Strategic planners, at both institutional and service levels, need to gain an understanding of how big data can be used to help planning, and also to work with IT services to define the semi-structured and unstructured data that should be collected. This will help to determine how it should be transformed in order to enable meaningful queries.
Many students are likely to benefit from learning analytics, with better information on how they are progressing in comparison to peers or previous cohorts and automated suggestions about how they can improve. They are also already being helped with pathway planning so that they can sign up for the modules in which they have the highest chances of success.
Learning analytics systems are already in place in some institutions.
Teaching staff will have better details of how their cohorts are progressing and receive prompts to intervene when students are at risk of dropping out or failure.
Nottingham Trent University: tutor dashboard
Nottingham Trent University piloted a tutor dashboard with a number of programmes. The system takes data from seven sources and determines what success looks like for the average student on each individual course. A visualisation chart shows a progress line that aggregates the seven data sources to indicate participation. An engagement rating is calculated for students and email alerts are sent to tutors when students fail an assignment or have had no engagement for a fortnight.
Most tutors use the dashboard once a week, which has minimal impact on their workload. Despite initial resistance from staff the system has proved popular and it has recently been rolled out across the institution.
Manchester Metropolitan University (MMU): curriculum refresh
Manchester Metropolitan University undertook a major curriculum refresh, enabling the creation of new course codes and other course-related tags that could then be used consistently across student records, the virtual learning environment (VLE) and timetabling. This was done to achieve better personalisation for the student, for instance to produce personalised timetables and assessment schedules in Moodle and the student smartphone app.
Meanwhile, as part of MMU’s day-to-day quality enhancement procedures, programme and module leaders are responsible for creating and tracking 'improvement plans', which have traditionally been paper-based.
Programme and module leaders are now provided with dashboards and tools to capture and track follow-up actions to support them in developing targeted improvement plans.
Bridgwater College: student portal
Bridgwater College deploys a student portal showing attendance and performance information when tutors meet with a student. If attendance is poor or a deadline has been missed emails or letters can be generated and sent to the student and/or their parent. Conversely, parents and students can be sent a letter or email when the student submits a good assignment.
The data is also available to managers at individual, programme and college levels so they can view outcomes against national comparators for attendance and retention.
Metrics include progression rates, success rates, destinations (eg university or employment), and where the college sits against national Level 3 value added data.
The college also gathers information from student surveys, some of which (for example satisfaction rates) can be benchmarked against external data.
King: interaction recording
King records all users interactions with its games in order to understand user behaviour and increase engagement and, because of its massive user base, it has been a pioneer of techniques in how to analyse data at scale quickly and efficiently to improve player experience (and therefore revenue).
Such techniques are transferable to education, for example users' engagement with e-learning tools.
We are undertaking a major programme to support the development of learning analytics in UK higher and further education which includes:
- A support network, including regular workshops and events (see our effective analytics blog or sign up to the [email protected] list for further details)
- A basic learning analytics system, available freely to universities and colleges, including a learning records warehouse, predictive analytics engine, staff dashboards, a student app and an alerts and interventions tool
- Support for institutions wishing to investigate their readiness for learning analytics and pilot the basic learning analytics tools
Support with legal and ethical issues
Legal and ethical issues in the use of student data for learning analytics will inevitably arise.
We have developed a code of practice for learning analytics to help institutions identify and deal with the main issues that arise.
During a discovery phase, institutions can receive support in assessing where they are and what they need to do in order to move forward with learning analytics.
We also provide assistance with piloting the basic analytics system.
Big data for business intelligence
We are also investing in services to allow institutions to exploit big data for business intelligence.
Activities include the joint development by the Higher Education Statistics Agency (HESA) and Jisc of the Heidi Plus and Analytics Labs services, investment in data visualisation tools (in particular Tableau) to allow exemplar visualisations, the development of new agile approaches to solving business problems with big data and joined up data sets, and working to bring data providers together to allow analysis across data sets.
Heidi Plus (HESA)
Designed to serve a wide range of staff roles and support improvements through visualisations and analysis tools.
Analytics Labs (Jisc)
A national initiative (formerly known as Heidi Lab) engaging with 115 experts from 72 higher education providers to identify new business questions, likely data and undertake analysis for new service content.
Cost management: Financial X-ray
Other Jisc services also have some bearing here, notably Financial X-ray, which helps IT departments to benchmark, understand and easily compare overall costs.
Financial X-ray is a useful tool to build a business case for changes in IT infrastructure, and create an ongoing mechanism for dialogue between finance and IT departments. It can also provide a means of highlighting the comparative cost of shared and commercial third party services.
Real-time decision support
The research and education sector is used to an annual reporting, decision-making and budgeting cycle centred on the academic year. This has served us well, but limits our ability to respond in an agile way to developing events.
There is huge potential gain from being able to respond quickly to an emerging opportunity, or to proactively address a developing problem. For example: in an over-recruitment situation for a science, technology, engineering and mathematics (STEM) course with intensive lab sessions, being able to re-arrange both the teaching schedule and lab supplies to ensure that all students in the cohort are able to participate fully.
This is an area where we could work with institutions using the learning analytics and business intelligence services to explore what a holistic approach to Gartner's 'digital education moments' might look like.
Efficiency and value for money
The Financial X-ray tool focuses on costs of IT service provision, but has potential transferability to other institutional activities where significant capital or operating expenditure occurs.
This approach could be particularly attractive in cases where institutions are looking to work together more closely, eg through collaborations such as the N8 Research Partnership or further education colleges looking to merge operations as a result of the area review process.
Groups such as the regional purchasing consortia, Colleges Finance Directors Group (CFDG) and British Universities Finance Directors Group (BUFDG) would be natural partners in this work.
Towards an integrated data innovation strategy
To enable our sector to exploit its data fully and to drive efficiency and productivity, we need to understand more about where core datasets live and how best to integrate them with new systems such as the national learning analytics and business intelligence services.
Developing these services has shown us that in many cases key information is locked away in proprietary or poorly documented systems and services.
The Universities and Colleges Information Systems Association (UCISA) annual survey of corporate information systems shows us the products that are in widespread use. Further work would be required to document application programming interfaces (APIs) and database schemas and develop convenient data extraction and integration tools - an 'education API' by any other name.
The From Bricks to Clicks report from the Higher Education Commission (HEC) highlighted the need for an integrated data innovation strategy for UK higher education. Significant progress has been made through the Higher Education Data and Information Improvement Programme (HEDIIP) coordinated by HESA, and this agenda is now being taken forward by Jisc, Universities UK, HESA and our M5 Group partners at the Quality Assurance Agency (QAA).
We believe that there is also huge untapped potential for the further education community to benefit from a similar approach.
Two of the areas we have touched on in this report are substantial topics in themselves, and will be the subject of future Jisc horizon scan reports.
Personalised and adaptive learning
Moving beyond discredited theories of learning styles to find ways of using technology to help each learner to progress according to their own needs and capabilities. We will consider how a personalised learning approach can be used to enhance the student experience.
The Internet of Things6 and the digital institution
Sensors and connectivity are increasingly being added to devices as diverse as forks and toothbrushes, door locks and even smart tattoos.
We will look at how institutions can exploit this trend and how Nesta's personal information management services could give learners control over who is able to do what with their personal data.
We hope you find our report useful and we would love to hear your thoughts about how we should take the conversation around data-driven decision-making forward.
Please do get in touch to discuss it further using the contact details below.
- 1 Now known as Analytics Labs
- 2 XML explained on Wikipedia - https://en.wikipedia.org/wiki/XML
- 3 NoSQL explained on Wikipedia - https://en.wikipedia.org/wiki/NoSQL
- 4 Map Reduce explained on Wikipedia - https://en.wikipedia.org/wiki/MapReduce
- 5 Data lake concept explained on Wikipedia - https://en.wikipedia.org/wiki/Data_lake
- 6 Internet of things explained on Wikipedia - https://en.wikipedia.org/wiki/Internet_of_Things