This document discusses using data mining techniques to build process models from full-scale plant data to optimize water and wastewater treatment processes. It provides several case studies where neural networks were used to model relationships between key process variables and contaminant levels. For example, one case study showed turbidity, color, and temperature accounted for 74% of the variability in chloroform levels. The document recommends using process models to predict contaminant levels, optimize chemical dosing, and evaluate "what if" scenarios to reduce operating costs while meeting regulations.