Data Enrichment Key to Enhancing Accuracy of AI Models in Fintech

Data Enrichment Key to Enhancing Accuracy of AI Models in Fintech

by September 14, 2022

Data enrichment, the process of enhancing internal data with relevant, contextual data obtained from external sources, is critical for financial services companies looking to get the most out of their investments in artificial intelligence (AI), allowing them to build more accurate predictive models and improve decision-making, says Mobilewalla, a Singapore-based consumer intelligence solutions provider.

In a new paper titled Improving Predictive Modeling Accuracy for Fintechs with Data-Centric AI, the firm explores why data quality, breadth, and depth are crucial for businesses to build accurate predictive models, and how data enrichment and feature engineering benefit AI in fintech.

According to the paper, while the majority of attention related to AI concentrates on complex ML techniques and refining algorithm code, it is critical for financial service providers to remember that the data used to train algorithms can be even more impactful to predict modeling accuracy.

The paper cites credit rating as a use case where information collected directly from applicants is often insufficient to filter out likely defaulters and prevent fraud. Instead, data collected from applicants should be enriched with additional information like location, demographics and behavior patterns, and more, to enable a more accurate credit assessment, the paper says.

These statements echo those made earlier this year by Mobilewalla founder, CEO and chairman Anindya Datta. During a Fintech Fireside Asia panel discussion hosted by Fintech News Singapore, Anindya said that while some information, like household characteristics and app engagement, may appear worthless in assessing one’s propensity to default, they are actually predictive of loan default likelihood.

More than a dozen buy now, pay later (BNPL) players rely on Mobilewalla’s data to assess consumer default risk as well as in the debt collection process, he said, noting that their growth and success have partly derived from their ability to make use of alternative data to assess risk, ultimately expanding access to credit to those lacking traditional credit data.

image via freepik

Mobilewalla, a leader in consumer intelligence, collects, cleans and processes a rich dataset, which can then be used by enterprises to better understand their customers. In the finance sector, the company has worked with the likes of Kredivo, Indonesia’s top BNPL brand, allowing them to segment their customers more appropriately, tailor customer experience and cross-sell other digital solutions post acquisition.

Rising demand for third-party data and data enrichment techniques in the finance sector comes on the back of booming adoption of AI in the industry.