Data science has been on the lips of almost everyone in business for some years, with executives asking how they use data to better understand their customers, competitors and the marketplace.
The answer was: by hiring data scientists and teams of developers. MINERVA, the new cloud-based software suite from Analytics Engines, intends to change that.
Founded in 2008 as a spin-out from Queen’s University Belfast, Analytics Engines initially worked on bespoke analytics solutions, with a customer base including RTÉ, the National Gallery in London and British government agency Innovate UK.
“A theme in all of this was using data analytics capabilities,” said Brendan McCarthy, chief commercial officer.
“In the National Gallery we looked at how visitors were moving through the gallery, how they were using dwell time, and providing insights into how visitors use that estate. That was something that allowed them to validate what they may have [previously just] thought. We provided the hard data behind that and did it in an intuitive visual manner,” he said.
For RTÉ, Analytics Engines aggregated and ingested multiple public data sources and then used complex multi-modal search criteria to make sense of it.
MINERVA is Analytics Engines’ attempt to push this previously bespoke technology out into the wider marketplace.
“It’s a new product that we are bringing out, as we've seen a few gaps in the market,” McCarthy said.
The key to what it does is give domain-specific experts the ability to visually comprehend data without needing skills in data science itself.
Typical users are to include financial services professionals and solicitors.
“Where we would see it is in the enterprise space, where you have domain experts in their field. Common among all enterprises is the question coming from c-level, ‘We’ve got all of this data so how can we use it, as we’ve heard that data is the new oil?
“I say, ‘data is the new oil – but who’s going to refine it for you?‘,” he said.
MINERVA can bring together internal data and enrich it with open data sets such as information from the land registry, Companies House and the electoral roll.
“If you’re working off data only from within the bounds of your company then you’re missing a richer picture,” he said.
This aggregated data is then examined by the financial and legal professionals who need it, using an intuitive visual and search-based interface.
“Historically, that would be done not just by data scientists, but also by engineers writing specialist code. We’ve taken that and now it can run from the cloud. A domain expert, who understands what's important to your business, can do it,” said McCarthy.
Concepts that we all use day-and-daily are used to sort data, meaning with a minimum of training seas of data can be brought under control.
Key to this is not ‘cleaning’ public data, as many working with data currently do, because in cleaning data you strip out a lot of ancillary information.
“If you do that then you lose history along the way,” McCarthy said.
“To us that history is data; you just haven't figured out how to make sense of it. What we’re trying to do is show you a relationship, and then you can try to start and figure out what is happening.”
Risk in business
A key area for Analytics Engines is risk. For example, its ability to examine Companies House data, rendering it quite literally visible, has obvious uses in enhanced due diligence, know-your-customer and fraud detection.
“Let’s say a customer signs up to a regulated service. They have to sign up to know-your-customer (KYC). Well, that’s a one-time event, but if you run it three months down the road: would that regulated customer be interested to know that they now have four degrees of separation from a known money launderer? What does that mean from a reputation or monetary exposure risk?”
For McCarthy, this represents an entirely new way of looking at data: hidden relationships can be revealed, and, along with them, risk.
“We believe that as people get past the rules-driven ‘a equals b equals c’ to embrace that fuzziness, saying ‘Okay, it’s got complex now and it’s gone past all the orderly data’, you can get rid of the rules and look at the [real world] linkages,” he said.
“I can’t get that visualisation of the relationship if I strip the data down to be completely clean.”
From equities traders to consumer banks seeking to combat so-called ‘money mules’, the possibility for reducing risk is enormous when relationships are laid bare.
“It’s not about the individual, it’s about their network,” said McCarthy.
“Financial services and that broader pool, anyone who has that KYC requirement, the investment profession, the legal profession, they can all benefit from this. The other is sales relationships, and this would leverage both internal and external data, showing you which businesses are close to current clients.
“If you’re a bank, do you want to know about your customer? We think that you do,” he said.
The ease of use of the visual and search-based interface makes MINERVA easy to use, but more advanced users can still dig deeper should they want to.
“We’re very cognisant that there are going to be different user groups. You might have a large group of people who do it in a 'no-code' way, but they can bring in their data scientist who can develop code if it’s needed for special cases,” McCarthy.
This seeks to democratise usability across a wider user base but also frees-up the scientist to do the specialist, difficult things, he said.
“That data specialist is not spending all day doing the generalist stuff, so they’re more productive as they aren’t doing the day-to-day stuff.
“We’re not getting rid of data scientists; we’re moving data from being a specialist IT domain to a real business tool,” he said.
Thinking machines mull risk
Risk and reward go hand in hand: ask any gambler. Most business executives acknowledge this but unlike casino-dwellers they prefer to take a more rational approach to risk than just rolling the dice.
In the past this may have led to an overabundance of caution and, in turn, lower returns. Increasingly, though, data is seen as a way of squaring the circle, including by using artificial intelligence to better understand risk.
According to a 2019 report by analysts Gartner, more and more businesses want to use data to drive decisions. Seventy per cent of chief auditors are more likely to act on recommendations based on data-driven insights, it found.
Getting the right people to turn raw data into meaningful information is not easy, though: in the same year, Britain’s Royal Society noted that demand for data scientists and data engineers rose by 1,287 per cent and 452 per cent respectively, compared with an increase of just 36 per cent for all types of workers during the same period.
Some industries will have a greater appetite for risk than others – insurance, for example, is all about risk – but even insurers are now looking beyond actuarial tables in order to understand risk on both a micro and a macro level.
Dr Sarah Bourke, chief executive of space software company Skytek, said that her experience of developing Skytek’s REACT software shows that data and artificial intelligence have a real contribution to make to insurance.
“AI and other intelligent systems are set to revolutionise insurance. Remember, insurance is all about risk: without risk there would be no insurance needed and, as a result, no industry. But insurers don't want to walk blindly into risk,” she said.
In the end, data is a trail left by activity, whether left by humans or natural events, and understanding it means getting a clearer risk profile.
“In marine insurance specifically, weather and climate data is being used to avoid total loss situations for vessels and installations, but there is more to come,” said Bourke.
“Data sets from ports on maintenance, crewing numbers, ownership registries: all of this can be fed into the decision-making process by using AI and machine learning to sift through the noise and recognise patterns.
“Questions can be raised around like sanctions breaches and other ‘dark activity’ or ships unprepared for rough weather. With insurers using data, it means there is a requirement for vessels and owners to eliminate risky activity.”