A meta heuristic optimization algorithm (Lion-BES-XGB) for water quality prediction

ABSTRACT Recently, water contamination has become a major problem in developing countries due to urbanization and population growth, leading to an increase in morbidity and mortality rates.Therefore, accurate water quality prediction is crucial in the urban water supply system. In this work, we developed a prediction model based on Extreme Gradient Boosting (XGB) using a hybrid feature selection approach combining Lion Swarm Optimization (LSO) and Bald Eagle Search (BES). The proposed method LSO-BES-XGB consists of three steps: preprocessing, feature selection, and classification.Z-score normalization helps fill in missing data values by scaling to indicate the number of standard deviations from the mean. LSO-BES Feature selection identifies relevant features, and the XGB classifier determines whether the water is normal or contaminated. The LSO-BES-XGB model was applied to the Cauvery River data set and achieved 94.22% accuracy, 93.12% precision, 94.23% recall, and 92.45%.