Approximately 4.48 million radon measurements, along with 186 radon-related contributing factors, were used to train a random forest algorithm to extrapolate zip-code-based community risk.