What This Competition Teaches About Better Understanding Your Data

Ready, set, go. Equifax data scientists around the globe sprung into action. The project was to investigate how to classify bank transaction data to make it more meaningful. Could it be categorised in a way that would enrich insights and help lenders better assess risk, ultimately improving customer experience? Competitors were furnished with raw unstructured transaction data from Spain and told to use advanced analytics and innovative techniques to solve this challenge.

With the roll-out of Open Banking in Australia and the opening up of new data sources, it’s a competition with significant real-world application. In today’s connected world, as companies sit on an increasing amount of data, many struggle with how to apply data to solve critical business needs. Those that are successful, use data in a way that elevates their business. Leveraging the insights in data to better target customers and discover opportunities to upsell and cross-sell products and services across the entire customer lifecycle. For financial institutions, understanding income and expenditure in greater detail offers opportunities to improve credit risk decisions and acquire new customers who fit specific criteria. Thus, striking the right balance between risk and growth.

How the Aussies innovate

Data scientists in this challenge knew that to categorise the transaction data to enable better risk decisions, they had to find the most predictive way to understand the text data. Their system needed to quickly and accurately ascertain whether text like ‘ABC Burgers’ should be categorised as an eating out expense. Or whether the text ‘Hair Cuts r Us’ is indicative of discretionary spending on personal care, and so on.

The Australian team decided to tackle the challenge by using Natural Language Processing (NLP) alongside machine learning. A branch of artificial intelligence, NLP aids computers to decipher and make sense of human languages. It involves the application of algorithms to identify natural language rules, enabling unstructured data to be understood by computers. 

The Aussie team hoped to achieve greater accuracy of insights by applying machine learning to data categorisation rather than taking a rule-based approach. They knew as the machine learning algorithms gained experience with more data fed in, the better the results, the higher the efficiency. This continuous learning and evolution would allow the algorithms to make more accurate predictions, faster. And that’s exactly what happened. 

The Aussie team knew not a word of Spanish. Yet, they were able to achieve accuracy rates of more than 90% on data which they couldn’t even read themselves, delivering a solution that is scalable to any language in the world. All this was achieved by a small team using the Equifax Ignite^® platform, a data analytics powerhouse capable of performing reliable, comprehensive analyses on all types of data.

Using your data to grow – key takeaways from an Analyst

This project demonstrated that given the right technology, applications and methodology, it’s possible to create meaningful insights from your data no matter what its source. “It’s your blue sky,” says Principal Analytics Designer, Dr Carlos Leung, who was part of the Australian team. “You might be sitting on data you think might not be very valuable. When, in fact, it has the potential to provide a lot of insights to your business.” 

Here Leung shares his key learnings from the project:

1. Size doesn’t matter

It’s a common misconception that companies need enormous resources and large teams of analysts if they are to mine their data effectively. Yet this project, with its five-person team of analysts, showed this to be an outdated belief. The technology and applications now exist for any business, regardless of scale, to find meaningful insights in data. 

2. It’s not the infrastructure that counts

Another false belief is that the only way to scale the bandwidth of your current analytical platform is to undergo a complex, expensive infrastructure upgrade. This project showed that by using the cloud-based computing capacity of Equifax Ignite^®, it’s possible to manage mass amounts of data, scaling up or down as required. Insights are generated rapidly using techniques like machine learning and artificial intelligence to sift through large volumes of disparate data across multiple sources. 

Ignite^® showed itself to have the power and capacity to store and process a large dataset while simultaneously interacting with it – enabling complex data analysis to occur in minutes or hours. This level of automation is critical to effectively analysing vast amounts of unstructured data.

3. Knowledge is power

Data scientists involved in this project looked to their peers from different countries and jurisdictions for ideas on how to progress their knowledge. So too, businesses can achieve results they never thought possible by tapping into professional expertise around big data and analytics. 

With a well-connected team of data and analytic experts around the globe, Equifax solutions have the competitive advantage of being infused with the knowledge, expertise and insight of many talented minds.

4. The importance of Natural Language Processing is growing 

Wealth and lifestyle insights, employment and income data, third-party data – there is no end to the breadth of data sets that might contain valuable insights. As we move towards Open Banking, companies will be tapping into different sources of data from all different places, including accounts from other countries (as was the case with this project). 

To capture the untapped potential in alternative sources of data and transform it in a way that delivers value requires a flexible approach that doesn’t need constant adjustment. This is where Natural Language Processing is critical for businesses who want to go beyond the standard data sets used for risk, for customer management and for growing their business. 

Not only does NLP fill the gap between human communication and computer understanding, but it allows for a broad range of approaches for deciphering unstructured text. Technology platforms like Equifax Ignite^®, that are integrated with NLP, provide countless opportunities for companies to revolutionise how they approach new and existing problems.

Accelerate your access to insights and make better decisions with Equifax Ignite^®. See the power of Ignite^® for yourself, complete the Request a Demo form.

Request a demo