Using the IDA-developed machine learning tool called Retention Prediction Model (RPM), we provide the Air National Guard (ANG) with enhanced training demand forecasts. Th RPM uses Defense Manpower Data Center data and ANG training data, spanning 2005-2020 and comprising about 1.4 million person-years for about 215,000 enlisted airmen. Results show the model correctly classifies 71% of service exits and 79% of service retentions across the period.