Browse by Type

Report #301 | | Members | Sign In

Big Data and Credit Unions: Machine Learning in Member Transactions

Credit unions can take a cue from big corporations by using transactional data to mine for useful insights around key business questions. Big data used right can help improve underwriting, predict members’ next products, and help members build wealth.

Executive Summary

Companies as varied as Amazon, Google, Walmart, and Wells Fargo are turning to “big data” for customer insights that will help them serve clients and capture market share. Big data is the analysis of huge data sets, and while individual credit unions may not have the resources of a corporate giant, advances in data storage and software tools mean that credit unions can start using similar tools and deriving similar value.

What is the research about?

For this research, five credit unions in the United States and Canada proffered their members’ anonymous profile information and transaction details to the researcher, who used variables as diverse as gender, product balances, credit score, income, and transaction amounts to search for revealing correlations. The findings show that some simple patterns evolve using big data and machine learning (a branch of artificial intelligence that focuses on the construction and study of systems that can learn from data). In particular, we found that members follow simple paths during their life cycle and adopt different consumer products at each stage.

What are the credit union implications?

Machine learning is a branch of artificial intelligence that focuses on the construction and study of systems that can learn from data. Because a machine learning project is only as helpful as the data that flow into it, the participating credit unions got the most specific insights. But their combined data still offer generalizable findings for all credit unions.

Simply having a 100-­petabyte cache like Facebook or 100 gigabytes like a typical credit union doesn’t guarantee any insights. Credit unions are best served when they start with a goal and only then decide whether machine learning can get them there. But don’t sit this one out. Big data is here to stay.