Consumer profile evaluation in a water distribution network using clustering and classification methods

Improving consumer profile evaluation is essential for effective water resource management, by creating a more accurate overview of the water distribution network. When compared to the classification by households, a data-driven approach can deliver more accurate results about the consumer types according to their behavior. Clustering methods are compared with classification models for evaluating the consumer profiles and their level of similarity with the consumer classes. The proposed approach evaluates the level of similarity between the initial classification and the consumer profiles obtained from the actual consumption data, using a publicly available dataset.