Data Sheet 1_Prediction of chemical oxygen demand in industry effluents using machine learning and IoT: a case study in Tequila, Jalisco.csv

The tequila industry in Mexico generates large volumes of wastewater with high organic loads, making real-time monitoring of chemical oxygen demand (COD) essential for regulatory compliance and environmental protection. This study presents a case study employing internet of things (IoT) sensors and machine learning (ML) algorithms to predict COD from suspended solids, dissolved oxygen, turbidity and electrical conductivity. The data was collected over an 87-day period, with 4,038 records obtained and employed in model development and validation. Three ML models (random forest, XGBoost, and gradient boosting) were evaluated using R2, where gradient boosting yielded the best results (R2 = 0.9878). Results indicate that while the three models exhibit good accuracy (R2 > 0.95) and do not show signs of overfitting (explained using residual analysis) they struggle predicting extreme (i.e., exceedingly low or high) values. Additional analysis were conducted to ensure model robustness, and residuals exhibiting homoscedasticity and approximate normality. This highlights that integration of IoT and ML offers a scalable and cost-effective solution for real-time water quality monitoring.