Navigating financial turbulence: How Mid-Hudson Valley Federal Credit Union used AWS AI/ML to enhance forecasting and decision-making

Mid-Hudson Valley Federal Credit Union (MHVFCU) is a member-owned financial cooperative serving individuals and businesses in New York’s Hudson Valley region. Founded in 1963 to serve IBM employees, it has since expanded its membership eligibility to include residents, employees, and businesses in several counties in the area. MHVFCU offers a wide range of financial products and services, including savings and checking accounts, loans, credit cards, mortgages, and investment options. As a credit union, it’s a nonprofit, focusing on providing competitive rates and personalized service to its members rather than maximizing profits for shareholders.

Starting in late 2022, the banking industry faced unprecedented challenges due to a rapidly changing economy. The inverted yield curve and rising interest rates created a volatile environment where many financial institutions experienced significant deposit runoff.

MHVFCU was no exception to this industry-wide trend. Their leadership team recognized the need for a comprehensive solution to closely monitor these changes and consolidate vital information in one place, enabling them to navigate this tumultuous financial market more effectively.

The credit union’s existing tools, including those used for data analytics and visualization, were outmoded. The credit union needed to automate their tools and develop advanced forecasting capabilities to make informed decisions for deposit management and growth to maximize benefit for their members. For this purpose, MHVFCU leadership sought a solution that would not only save time in data preparation and analysis but also provide accurate forecasts for balances across four key deposit products: savings, checking, certificate deposits (CDs), and money market.

Solution

MHVFCU chose to work with Amazon Web Services (AWS) due to the powerful data analytics and visualization capabilities offered by Amazon QuickSight and the robust artificial intelligence and machine learning (AI/ML) capabilities of Amazon SageMaker and related services. The credit union was particularly interested in using Amazon SageMaker Canvas to expedite building and deploying ML models without extensive coding, giving them the flexibility to iterate and improve their balance forecasting capabilities.

The following diagram shows the solution architecture.

Figure 1. High-level architecture for MHVFCU’s new predictive analytics workflow using Amazon SageMaker Canvas

MHVFCU implemented an end-to-end solution using several AWS services including AWS Glue, and Amazon Simple Storage Service (Amazon S3) for data ingestion, Amazon SageMaker Canvas for model development, and Amazon QuickSight for data visualization and business intelligence (BI). The solution architecture allowed MHVFCU to:

• Ingest data seamlessly from existing holdings in Snowflake.
• Use AWS Glue for extraction, transformation, and loading (ETL) processes.
• Use SageMaker Canvas for advanced forecasting.
• Enhance the use of Amazon QuickSight to derive business insights.

Results and benefits

By implementing this AWS-powered solution, MHVFCU realized several significant benefits:

• Time savings: The solution reduced data preparation time for the accounting team by approximately 15 minutes per day. QuickSight minimized back-and-forth communications and reduced one-time questions, saving variable amounts of time across teams.
• Enhanced forecasting capabilities: MHVFCU gained the ability to monitor and evaluate balances across four key deposit products with greater speed.
• Improved decision-making process: Senior leadership receives weekly reports to derive business insights, a process that was previously performed manually.
• Scalability and flexibility: The solution allows for expansion to new use cases, such as analyzing the frequency of nonperforming loans in their payments division and experimentation with related time-series forecasting to predict the impact of interest rates in account balances.

Looking ahead

Eugene Corcione, Director of Data Analytics at MHVFCU, commented: “The communication and support from AWS have been great throughout this process. We’ve learned valuable lessons about the importance of data preparation and creating comprehensive data dictionaries to identify crucial information.”

MHVFCU is continuing to refine their use of AWS AI/ML tools and is exploring additional use cases. By using AWS AI/ML services, Mid-Hudson Valley Federal Credit Union has positioned itself to make data-driven decisions more efficiently, ultimately leading to better service for its members and improved operational efficiency.

AWS is committed to supporting credit unions in their digital transformation. Our dedicated AWS Account Teams, specializing in credit union needs, work closely with institutions to help them maximize the benefits of our services.