Photo By Dori Whipple | Matt Luckey served as the lead for the Joint Munitions Command’s award-winning...... read more read more
ROCK ISLAND ARSENAL, Ill. — The Joint Munitions Command earned first place in the predictive and prescriptive analytics category during the 2025 Army Materiel Command Data Summit.
The AMC Data Summit brings together analytics professionals, PhDs, contractors, and internal talent from across the Department of Defense to share challenges, successes, and forward-looking priorities. A highlight of the event is the Data Showcase competition, where teams present their most innovative projects across various analytical competencies.
JMC showcased a project that demonstrated not only technical excellence but also strategic value to the broader enterprise. JMC’s winning submission addressed the challenge of short-term and long-term financial planning. By applying predictive and prescriptive analytics, JMC’s team delivered measurable insights into likely future events, helping stakeholders at all levels make faster, more confident decisions. The goal was to build a platform that could forecast future funding, adapt to new information, and communicate insights clearly and effectively.
“We focused on telling the right story to the right audience, through a simple, transparent interface,” said Matt Luckey, the project lead for JMC. “That clarity was just as important as the analytics behind it.
“Given the complexity and visibility of the effort, the team adopted a hybrid project management model, blending agile and waterfall principles,” Luckey added. “This approach allowed for frequent feedback and rapid iteration while still meeting leadership’s documentation and updating requirements.”
At the core of JMC’s project was an ensemble forecasting model, designed to improve accuracy and transparency of the approach. Members of the JMC team tested several forecasting techniques to create a baseline and evaluated them using the Mean Absolute Percentage Error, Root Mean Squared Error, Mean Absolute Error, and tracking signal metrics.
Building on this foundation, JMC integrated open-source machine learning models, ultimately selecting a Bayesian inference approach that could adapt dynamically as new data became available. Forecast performance was further enhanced using optimization tools like Excel Solver and nonlinear programming, allowing for fine-tuning of model weights and exploring funding scenarios.
A standout feature of the solution was the use of a large language model (LLM) to process and extract insights from unstructured funding documents. The LLM identified potential risks associated with unfunded requirements, producing concise, actionable outputs for each funding category. This automation replaced hours of manual review with seconds of analysis, further accelerating and improving the budget planning process.
“The final product was more than a model; it was a centralized, interactive decision-support tool aligned to mission outcomes,” said Marty Carlin, Chief of Data and Analytic Operations for JMC. “Whether a user was a financial planner, manager, or senior leader, they could access relevant information at the right level of detail, delivered in a format they could act on.
“This project now serves as a blueprint for what’s possible when advanced analytics, storytelling, and operational context come together,” he added. “As JMC continues to expand its analytics capabilities, the project stands as a reminder of how to translate complexity into clarity, and analytics into action.”
| Date Taken: | 08.12.2025 |
| Date Posted: | 08.12.2025 08:33 |
| Story ID: | 545388 |
| Location: | ROCK ISLAND ARSENAL, ILLINOIS, US |
| Web Views: | 19 |
| Downloads: | 0 |
This work, For predictive and prescriptive analytics, JMC wins first place at AMC Data Summit, by Matthew Wheaton, identified by DVIDS, must comply with the restrictions shown on https://www.dvidshub.net/about/copyright.
No keywords found.
