Statistical Modeling of Lead Service Line Locations Using Open‐Source Data

Abstract

Key TakeawaysLead service lines in drinking water systems pose a serious public health threat, but many utilities lack complete records of their locations.Statistical models trained from freely available public data sources have the potential to provide an efficient way to prioritize service line inspection plans.Factors such as choice of machine learning algorithm and spatial distribution of parcel records must be considered to ensure precision when developing a predictive model.