Amenity Counts Significantly Improve Water Consumption Predictions

Research by: Damian Dailisan, Marissa Liponhay, Christian Alis, and Christopher Monterola

EXECUTIVE SUMMARY

Changes in water demand in urban areas depend on several factors such as daily human movement, population size, land use, and amenity types among others. Mobility data derived from phones can capture human movement but is difficult to obtain and lacks the intended destinations of users. Here we present a framework for predicting water consumption in areas with established service water connections and generalize it to underserved areas.

Our work used features such as geography, population, and domestic consumption ratio to train machine learning algorithms that predict water consumption. We used 44 months of monthly water consumption data from January 2018 to July 2021, aggregated across 1790 district metering areas (DMAs) in the east service zone of Metro Manila. Results show that amenity counts reduced the mean absolute error (MAE) of predictions by 1,440 m3/month or as much as 5.73% compared to just using population and topology features.

Our models work even amidst the pandemic. Predicted consumption during the pandemic also improved by as much as 1,447 m3/month or nearly 16% compared to just using population and topology features. The robustness of the model to disruptions in human mobility, such as lockdowns, indicate that including amenities as a machine learning feature improves water consumption predictions. Our models could help water concessionaires in service planning as well as in bidding for new franchise areas under an increasingly competitive and regulated environment.

Keywords: machine learning, human mobility, pandemics, forecasting, decision tree learning, water resources

To cite this article: Dailisan, D., Liponhay, M., Alis, C., & Monterola, C. (2022). Amenity counts significantly improve water consumption predictions. PLoS ONE, 17(3), e0265771. https://doi.org/10.1371/journal.pone.0265771

To access this article: https://doi.org/10.1371/journal.pone.0265771

About the journal

PLOS One is a peer-reviewed open access scientific journal published by the Public Library of Science (PLOS) since 2006. The journal covers primary research from any discipline within science and medicine. PLOS ONE editors evaluate research on the basis of scientific validity, rigorous methodology, and high ethical standards, with the aim of making all well-conducted research freely available.

Journal ranking