Analytics for the greater good

Image credit: Pixabay

Positive change and interventions rely on good governance. Injustices can emerge organically and often unpredictably, but it’s when we do nothing that they’re allowed to grow and spread. For years we’ve depended on frontline services and response units to identify, resolve and prevent wrongdoing. This encapsulates everything from police forces tackling illegal drug use, social services rescuing vulnerable children from their abusers, to fraud teams exposing corruption. 

At some point in each of these processes, a decision has to be made before action can be taken. The most valuable quality a public servant can have, therefore, is good judgement. This boils down to being able to make the best-possible decisions with the insight available. However, in today’s non-stop digital world, it’s unfair and unreasonable to expect humans to make those decisions unaided.

Technology is a powerful tool for our emergency and public services. Far from trying to replace them, analytics augments the abilities of investigators and frontline practitioners, helping them to make faster, better decisions for the public. It gives them valuable insights from lots of information, so they can make the best possible decisions while drawing on their own expertise as well.

Clarity under pressure

We’re all bombarded daily with massive amounts of information. Any investigator – whether they work for the police, social services or an anti-fraud team – typically has mountains of both structured and unstructured data they must assess  before a decision can be made. 

The stakes can be high, so investigators need to be confident in the quality and accuracy of their insight. Yet, the best decision is hard to guarantee when timescales are short. When lives are at risk, the reality is that humans will not be able to explore every lead. Inevitably, connections will be missed and opportunities lost. 

Under pressure and under-resourced, decisions can be made mostly on gut feeling and experience. While experience is invaluable to an investigator,it will not lead to the best decision every time. Good decisions cannot be guaranteed without fast and valuable insights from analytics.

Image credit: Pixabay

Image credit: Pixabay (Image credit: Pixabay)

To ensure data is accurate and is being used to its best potential, analytics should be paired with AI or machine learning technologies. A solution that’s been trained on historic data or previous best practice knows what someone should be looking for. It can automatically make the link between seemingly disparate data sources, recognising patterns across masses of data that might otherwise have been missed. 

A trained AI system can then decide what connections are most important, helping the investigator prioritise their work. Furthermore, machine learning means the model can adapt itself as it encounters changes in the data.

In a field like tax compliance, where investigators must build up large bodies of evidence across many data sources, these capabilities are game changing. Tax fraudsters are adept at covering their tracks, using a multitude of different companies in various jurisdictions to hide asset ownership, profits and transactions. An analytics solution can bring all this information together and reveal the incriminating untruths that link them. Through a visual interface, this information can easily be communicated to the investigator, giving them the clarity they need to make the right decision. 

One company that is already ahead of the curve is Allianz Insurance. Using a hybrid, analytics-led approach to fraud detection, the insurer is able to sift through immense quantities of data, revealing organised fraud networks and communicating the information quickly and simply to investigators.  

Analytics on the frontline

Yet analytics isn’t just a useful tool for investigators in the back office. Increasingly, we’re seeing these technologies applied in scenarios where practitioners have only a split-second to react or make a decision. For example, monitoring traffic to ensure that vital emergency services are best directed to their destination. Insight is no less important in such a scenario, but practitioners need faster response times than those offered by traditional solutions.   

On the ground operations increasingly depend on Internet of Things (IoT) devices – like mobile phones and connected cameras – as a source of data and insight. However, many practitioners don’t have the luxury to wait for the data to be sent back to an analytics centre for processing. Event Stream Processing solves this challenge. This is the process of quickly analysing time-based data as it is created and before it is stored. By performing analytics closer to the source, results can be pushed immediately to those who need them.

Event Stream Processing combined with computer vision (a branch of AI) could help with maintenance of a railway line, for example. A drone camera could use its computer vision capability to analyse images and identify any defects in the line, and furthermore which require attention first all the way down to some that may not need attending to at all. . The drone delivers precious information about the state of the railway line, and the resource needs now and over time to repair it. This information can also ensure that there is minimal disruption to services, as resources can be prioritised to where repairs are needed most.       

Analytics and AI solutions can help our public services gather more data and deliver better insight. By automating certain tasks, they allow investigators and frontline practitioners to have more time to review insights, make decisions and be more productive.

Hugo D’Ulisse, Technical Director, Public Sector for UK&I at SAS

Hugo D’Ulisse

Hugo D'Ulisse is technical director for SAS's public sector team, helping government organisations deploy advanced analytics.

He is passionate about the role data, analytics and technology have in accelerating data-driven innovation from predictive analytics, Artificial Intelligence and Machine Learning with potential civil and social benefit, particularly within Scotland, but also across the UK.