Water quality prediction using enhanced parrot optimization algorithm with ensemble learning techniques

Abstract

Water quality monitoring has become increasingly critical as urbanization and industrialization escalate global water pollution challenges, contributing to approximately 1.8 million deaths annually from waterborne diseases according to World Health Organization estimates. This study introduces a novel integration of the Parrot Optimization Algorithm (POA) with ensemble learning (ANN, RF, XGBoost) for water quality prediction using the Cauvery River dataset (from 2018 to 2022, (642 samples, 26 parameters). The approach was validated using a comprehensive dataset from the Cauvery River collected by the Tamil Nadu Pollution Control Board. Experimental results demonstrate that POA significantly outperformed traditional optimization techniques, achieving 85.5% accuracy, 86.3% precision, 86.2% recall, and 87.3% F1 score. Crucially, we have statistically validated that POA exhibited 32% faster convergence and selected 50% more relevant features compared to Particle Swarm Optimization (PSO) and Genetic Algorithm (GA). This research contributes a computationally efficient and highly accurate decision-support tool that enables stakeholders and policymakers to monitor water pollution in real-time, facilitating early intervention strategies and promoting sustainable water resource management practices globally.