Data and code for Hybrid machine learning model for disinfectant dosing in small-scale water treatment under data scarcity

This repository contains all data, scripts, and supplementary materials associated with the research article entitled “Hybrid machine learning model for disinfectant dosing in small-scale water treatment under data scarcity”, submitted to the Journal of Water Process Engineering. The study presents a hybrid artificial intelligence approach that integrates Seasonal and Trend decomposition using Loess (STL) with Extreme Gradient Boosting (XGBoost) to predict optimal chlorine and fluoride doses in small-scale water treatment plants with incomplete historical data. The method was implemented and validated in a real-world facility located in Brazabrantes, Goiás, Brazil, achieving significant reductions in disinfection by-products (DBPs), chemical usage, and operational costs. This repository includes: Raw and processed datasets used for model training, validation, and testing; Python scripts for data preprocessing, STL decomposition, model training (ANN-MLP, RF, XGBoost), and performance evaluation; Visualizations and metric summaries used in the article (including Figures and Tables); Supplementary files containing extended methodological details and the WTP system description; A graphical abstract and final version of the manuscript in PDF format. All materials have been organized to ensure full reproducibility and transparency, in compliance with the data sharing policy of Journal of Water Process Engineering and aligned with FAIR principles. The repository enables other researchers, engineers, and decision-makers to replicate or adapt the methodology for use in other water treatment contexts.