Developing early warning systems to predict water lead levels in tap water for private systems

•Predictive models alert households about the risk of lead in drinking water.•Bayesian Belief Networks and Ensembles of Decision Trees classify lead in water above 15 ppb.•Models use data reported by at-home water quality testing kits to predict water lead levels.•Models are applied for a dataset collected at private drinking water systems in Virginia.•Experiments test the effects of household characteristics, location, and water quality data on model predictability.