Resource H.R. 1 Implementation

QC Data Workshop: Predictive Analytics for SNAP PER Reduction

This workshop summary synthesizes key takeaways from a convening of nearly 40 research and data analytics staff from 15 states focused on SNAP Quality Control (QC) data modeling.

Organizations: Georgetown University, McCourt School of Public Policy, Better Government Lab, Aspen Institute Financial Security Program, SCALE Lab, Yale University

Published Year: 2026

Download Resource

The resource outlines specific tools and predictive analytics strategies states are using to reduce SNAP payment error rates (PER), including machine learning algorithms like XGBoost and Random Forest.

It details helpful variables for identifying high-risk cases—such as shelter costs exceeding 50% of income and benefit amounts relative to maximum allotments—and discusses the importance of modeling agency and client errors separately.