Is the future of Education Analytics Open?

Recently I wrote a short positioning piece on ‘Free analytics’, and introduced the concept of Open Education Analytics (OEA).?OEA is an open-source community building tools to help make robust and impactful analytics within the education sector a reality.

I am now writing to evolve the thinking in the space further, and to put out a call to action not only to Multi Academy Trusts but to any education organisation in the UK implementing a digital strategy that includes analytics and artificial intelligence.??If you are an education organisation in the UK that is NOT implementing a digital strategy that includes these elements… I have a different but related call to action!

First off, if you are not familiar at a high level with OEA, please spend a few minutes to watch the overview video, and to have a quick look at the information on the website.

What’s the problem getting started with analytics in education?

At the outset, it’s the same problem as many other industries face.?Data is a bit like oil, in that in order to be valuable it has to be refined.?Over years, schools have created lots of data silos, using different systems to record different things.?So when it comes to finding the answers to questions such as:

“What factors should I focus on to improve attendance?”

“What impact does student wellbeing have on exam performance?”

?“Which students are exhibiting early warning signs of safeguarding risk?”

we find that the data we need is in the management information system, the safeguarding system, online student surveys, the assessment system and probably many more.

Orchestrating this data becomes an all-encompassing task – depriving us of time we should be spending implementing the interventions and change management to improve outcomes based on what it tells us.

The EdTech market currently only helps in a really limited way. Your MIS providers make some data available in reports, and via application programming interfaces (API) that lets those with sufficient resource fetch some data automatically.?You may also be using products like Groupcall Xporter, Assembly, Clever or Wonde to a similar effect.?

Beyond the MIS (Warp Factor 1)

Community Brands own Assembly Pro solution is leading the way providing analytics via the powerful Microsoft Power BI which automates a lot of the heavy lifting and is notable for bringing on board partners into the ecosystem such as CPOMS, SMID, and Welbee. Other vendors that offer more than one solution are increasingly building more integrated analytics across them.?But what priority do they put in building the interfaces to competing’ products??How long does it take companies in the market to build the relationships, align on the vision and then commit their investment dollar to making this a reality for the schools that need it??From experience, it takes months, and many times, years.

?I have been in school governance for many years and one of the things that drives the work that should be done is a sense of urgency.?In reality systems need to change to improve education outcomes, especially in safeguarding, children don’t have years and we have to intervene early.?They come this way only once.?The imperfect EdTech market is arguably doing its best in meeting the requirements of schools and the children they serve. However, it is not keeping pace with the speed of change in other industries already utilising the power of technology platforms being provided by Microsoft.

A utopian design

What if we were to design an education analytics system from scratch??

What would schools and MATs need?

We have discussed the need for unfettered access to all their data stored in different systems.?Of course, in our utopia, these systems have a common identity so that we can match schools and students in one system, with the schools and students in the others and therefore pull the data together.?

We are going to need a place to store all this data that is secure and when we use this data, even in very large volumes (such as attendance data) it must be performant.?We want reporting and analytics to be presented to different audiences in different ways.?Governors and Trustees need a summary, with the ability to drill into anonymous data as needed.?School leadership need access to data that is pertinent to them.?We would like to consider how we include all teachers, students, and parents in our analytics journey – to be transparent and offer them the insight on what might be done to improve.

We want to be able to move up the analytics ‘value chain’ – from descriptive to predictive to drive more value from our data:

Figure 1 – Analytics value chain

Our utopian system therefore needs to be able to easily build, train and run machine learning algorithms (part of artificial intelligence) on data at scale to provide these outcomes.?All of this, of course, needs to be conducted ethically – recognising privacy and fairness, being explainable and retaining human agency over the process.

We’d like to know how our patterns and trends compare, in real time, with that seen by other schooling systems regionally and nationally.?And because we are altruistic, we would love for the data generated to be available in the right form for research organisations to conduct research and positively influence national education policy.?We want to foster participation from educators everywhere to help determine what is important, to learn from others, and to make all this available tomorrow.

Our utopian system is also free.?Ok, so maybe not free, but affordable.?Sounds simple.

Back to Open Education Analytics

OEA provides the technology and a framework to support all of the above aims.?It does not, however, provide the human capacity to achieve this automatically.?If we want to move towards converting our utopia into a reality in the UK we are going to need to collectively apply some effort to make this happen.

Let’s address some practicalities.?Currently, under OEA, every MAT would have to create their own Azure Synapse (the tech bit that gets data, stores data, and delivers it out in analytics via Power BI), use the open-Source resources on Github (a free to use store of the technical resources) and build their own customisations.?Then maintain and support the environment over time.?Those with the cash to splash could hire a data engineer and a data analyst to do this, but the skills aren’t cheap.

Next up, the Github project currently contains modules (bits of code designed in the main to fetch data from a source system) for Microsoft 365 (great for getting student engagement stats) and separately licenced applications like Insights Premium (great for understanding Teams / online learning usage) or standards-based connectors to systems using SIF (an interoperability framework).?Other than that, modules available are quite specific or US focussed.?So any new UK entrant to OEA is going to have to start by building the modules to get at the data they need.

Figure 2 - Microsoft OEA Modules

There are of course great help videos and templates on how to go about this.?Given that many MATs will have similar requirements, rather than each start addressing this on their own, wouldn’t it be beneficial to have an active community around this actively discussing who may be doing what??Or in six months’ time will we find 3 SIMS connectors and 2 Arbor appearing, and no-one has done iTrent HR.

Of course, it would not make sense to start there in any event.?We would build the Xporter module (other school data extraction tools are available…) in order that it did not matter which MIS were in play and that all OEA participants would have an easy starting path on their OEA journey.

Share and share alike

If a community came together, we could build and fund a shared resource to build the modules that are important to the whole Education ecosystem and share in the open-source outputs of this activity.?This would achieve two things – first, the blocker of the EdTech market responding in a quick enough fashion to deliver against schools and MAT requirements would be removed, and secondly the prioritisation of what is to be done could ignore traditional company boundaries.?

A community approach could therefore solve a couple of these big barriers to adoption.?It could also support the commercial activity through working in partnership. While some large MATs would have the capability and capacity to pick up the open-source modules and implement them within their own Synapse environment, this will still be beyond 90% of MATs to accomplish.?At Community Brands, we are one of only a small handful of companies worldwide holding ‘Advanced OEA Partner’ status, and of course would be delighted to provide support to all those that needed it by hosting the technical aspects and managing the environments over time.?This removes the barrier for all, in a cost-effective way.

More than this though, if we got this right as a community with an approach that was ‘ethical-by-design’, we could start to build an environment at scale – but incrementally as MATs joined.?This source of data could be used for live benchmarking, for training predictive algorithms and conducting research in a manner that no single MAT would be able to achieve.

Even the largest MATs hosting their own Synapse and building their own customisations would be interested in live benchmarking and more robust data on which to train their own algorithms.

Figure 3 – The modern data platform architecture

The sector would definitely benefit by thinking more broadly than MATs.?Use cases here abound for Local Authorities, providers of assessment at scale and of course local and national government.?The ability to host this from an ‘Analytics-as-a-service’ (or AaaS – an acronym I love) perspective also offers us collectively the opportunity to use Power BI Premium capacity for all users at a reasonable cost – and moving away from the commercial limitations small and medium size MATs would otherwise find with affordability and licensing users individually.

With a commitment to open source, we could all be working towards the same vision to turn data into meaningful information to help us make better decisions and improve outcomes for young people.?That would be neat.

Call(s) to action

I’m supposed to do things in threes.?So here are three calls to action.

1.??????If a vendor does not provide free access to your data when you need it where you need it, in its most raw and transactional form, find a vendor that does.?As we move, together, towards sharing data and predictive analytics and AI you will be left behind if you don’t demand this.?Ask your vendors now for details on this – you don’t want to extract anything school by school, subject by subject, you want a big red button that says, ‘give me everything’ (and then one that says ‘give me everything that you’ve not already given me…’)

2.??????Don’t think about reports that you want to produce, think about the questions that you need answers to.?We need to stop wasting time pouring over report packs and more time being informed of the outliers and risks so we can focus on the next bit that computers are currently rubbish at and we need good old humans – “That graph is lovely, but what do I do about it?”.?I’ve some thoughts on how we go on to build machine intelligence for that bit too (prescriptive analytics) but we need to mature a bit through predictive first, and build better data.

3.??????Join us and the growing community of education professionals interested in how we move quicker, together, to build better analytics to help improve the outcomes and life chances of young people.?We’ll be putting more structure on what we should build, how people can contribute and what priorities should be as we move forward.?This is only going to work well if you engage and contribute too.