Segmentation and characterization of water and sanitation utilities in Mexico

Abstract

This study examines the structure and performance of Drinking Water and Sanitation Utilities in Mexico, using open data from INEGI and unsupervised Machine Learning techniques. A Principal Component Analysis (PCA) was conducted to identify the most representative operational variables, followed by the implementation of the K-Means clustering algorithm to group the utilities into homogeneous categories based on their structural and operational characteristics. The analysis resulted in seven distinct clusters. The first three clusters encompass 88% of the utilities, characterized by low revenues, limited infrastructure, and location in municipalities with very high, high, or medium levels of marginalization. These utilities exhibit average coverage rates ranging from 72.9 to 82.6 % for drinking water, and from 69.1 to 76.5 % for sewerage, with per capita annual revenues below $325 pesos. Clusters 4 and 5 demonstrate improvements in both coverage and operational efficiency, serving larger populations, with better infrastructure and per capita revenues reaching up to $916.96 pesos. Cluster 6, composed mainly of state-level utilities, shows coverage levels above 95% and per capita revenues twice as high as those in Cluster 5. Lastly, the Metropolitan Cluster (MC), consisting of 20 utilities located in large urban areas, presents the best performance indicators, with coverage rates exceeding 97% and average annual per capita revenues of $1 625 pesos. The results obtained allowed for the identification of common characteristics within each group, providing valuable information for the design of public policies and improvement strategies tailored to the specific conditions of each segment.