How is Data Science Reinventing Financial Services?

Following the second event ‘How Data Science is Reinventing Financial Services’ in our Fintech Series, Ed Adcock, Principal Data Science Consultant alongside Lokeshwari Mohan, Consultant and Geetanjali Sawant Senior Consultant Data Analyst provide their key takeaways looking at the foundations of data science, the technology behind it and how data science is changing the financial services landscape whilst delving deeper by outlining an ongoing case study.

What role does data play in financial services?

Whilst Data Science, as a discipline, focuses mainly on analytics and insights, the frameworks and technology used to perform data science and model building have a significant impact on approach, development and ultimately the delivery and results. The overall data space is key to understand the overall view of data in financial services (see fig 1).

• Data Strategy and target operating model – Frameworks to enable controlled implementation of processes and technology
• Data Technology – Technology and tools that facilitate the effective and efficient use of data
• Data Analytics – Modelling, data, applying algorithms and creating data insights based on data and modelling outcomes
• Data Insights – Utilising data insights to drive business decisions, optimize processes and deliver increased value from new and existing data assets

Using data science in the banking industry

We all know that whilst data science is not new, it had a slow start in the first few years but is now being heavily used across the banking industry – particularly over the past five years to solve complex challenges such as:

• Strategic Insights – Identify data linkages for better insights and decision making
• Personalization – Increase customer demand through personalised offerings
• Automation – The insights found can be used to optimise and improve productivity through automating processes

Embracing data science leads

By embracing data science across the business, influential change can be achieved through connecting all departments, countries, and regions across the business (mindful of local data regulation) and sharing the data to depict a holistic view of how the business is performing thus leading to enhanced strategic decision making from the smallest team to the largest global business.

The proliferation of No/Low-code platforms have the potential to dramatically increased developer productivity and empowers ‘citizen developers’, who guided by their domain expertise and an intuitive visual interface, can deliver insights and optimizations, without the need for specialist data science knowledge.

Businesses however have had mixed results with this citizen developer concept and poor consideration of overall objectives and training see initiatives being cut short due to burgeoning cost and unrealistic expectations and outcomes.

Managing risk through model decisioning

Taking a deep dive into a previous project, we developed and deployed an Automated Valuation Model (AVM) for prediction of house prices to support loan and mortgage applications. One key takeaway was the number of different technologies a Data Scientist uses to deliver just one solution.

This project is an example of managing risk through model decisioning. However, this does not replace the underwriter or risk manager, it only supports them to make their decisions after considering multiple models and data sources.

We also looked at the topic of model calibration – something that is fundamental to achieving consistency of measurement of an on-going model. It is best practice to be performed on regular basis or whenever the distribution shifts significantly.

Mapping the solution

The common issue across financial services is Duplication of entities. The effect of these duplications results in significant extra time and resource allocation during KYC due diligence processes. As millions of records exist in standard databases, data science solutions are well placed to find duplicates that couldn’t be filtered with an exact match.

The intelligent mapping solution performed analysis on millions of rows of data quickly and performed pre-processing logic to improve the accuracy of matches. Significant cost-savings were delivered to the financial institution and a new highly adaptable, widely applicable, and reusable tool, which is already being used for additional use cases.

How Delta Capita can help

Delta Capita’s experienced global data & technology team has wide-ranging knowledge across financial services and data science. With deep subject matter knowledge experts supported by data experts our delivery teams can adjust, guide, and deliver on complex data challenges facing our financial services clients in the face of increased regulatory scrutiny and Cost optimization.

To learn more, contact us here.

This article was written by Ed Adcock, Principal Data Science Consultant, Lokeshwari Mohan, Consultant and Geetanjali Sawant Senior Consultant Data Analyst.