Sign in Subscribe
Machine learning

Machine Learning Models Improve Drinking Water Quality Forecasting

Strategic Insights

— 1 min read

Researchers at the University of Girona, Spain have developed an innovative approach to enhance microbial water quality forecasting by combining accurate gradient-boosted decision trees (GBDT) predictions with well-calibrated uncertainty estimates. This study, funded by UdG, Spanish Government, and the Ministry of Science & Innovation, benchmarked state-of-the-art regression algorithms and uncertainty quantification methods for predicting E. coli concentrations in a drinking water catchment.

The research team, led by Hector Monclus, has successfully quantified uncertainty in E. coli ml-based monitoring, providing a more reliable approach to risk assessment and decision-making in drinking water management. The study has been peer-reviewed and published in the Journal of Water Process Engineering.

Key Takeaways:

  • Research focuses on improving management, decision-making, and monitoring of drinking water quality using machine learning regression models.
  • GBDT predictions proved effective for real-time tracking, with CatBoost achieving the lowest error (RMSLE = 0.877).
  • Conformalized Quantile Regression emerged as the most reliable method for uncertainty quantification.
  • Uncertainty estimates were successfully generated to identify high-risk contamination events.
  • This approach enhances microbial water quality forecasting, offering improved risk assessment and supporting robust decision-making in drinking water management.
  • Authors include Hector Monclus, David Abert-Fernandez, Ester Aguilera, Pere Emiliano, and Fernando Valero.

Statistics:

  • R ^{2} values for GBDT predictions ranged from 0.83 to 0.90.
  • RMSLE values for different regression algorithms were: CatBoost (0.877), Random Forest (1.150), and Naïve (1.160).
  • 5% improvement in error rate was observed using CatBoost compared to Random Forest.

Sources:

  • Beyond Point Predictions: Quantifying Uncertainty In E. Coli Ml-based Monitoring. Journal of Water Process Engineering, 2025;78.
  • NewsRx. Researchers from University of Girona Report Findings in Water Process Engineering (Beyond Point Predictions: Quantifying Uncertainty In E. Coli Ml-based Monitoring). Ecology, Environment & Conservation. October 17, 2025; p 1720.

Read more