A Robust XG-Boost Machine Learning Model for Water Quality Estimation System by Leveraging with Chi-Square Forward Sequential Feature Selection Technique

Water quality estimation entails evaluating multiple parameters encompassing physical factors and chemical constituents. Precise assessment of these parameters is crucial for safeguarding water resources, environment, and human health. In the context, machine learning is essential due to its ability to analyse complex datasets and identify intricate patterns within diverse water quality parameters. Machine learning models enhance accuracy, enable timely decision-making, and facilitate proactive interventions for effective water quality estimation. Towards this, the researchers introduce an innovative approach to water quality estimation utilizing a robust XG-Boost machine-learning model, augmented by the Chi-Square Forward Sequential Feature Selection technique. The proposed XG-Boost algorithm, renowned for its high accuracy and resilience, serves as the core predictive model. By integrating the selected features and fine-tuning hyper parameters, proposed XG-Boost model demonstrates superior performance in water quality estimation compared to Optimizable Naive Bayes, KNN and SVM methods. Rigorous Holdout cross-validation and evaluation metrics validate the proposed model's reliability and generalizability. Furthermore, the research addresses the challenge of model interpretability by elucidating the relative importance of selected features, offering a deeper understanding of the key factors influencing water quality. The results claimed in this study envisages, the proposed XG-Boost machine-learning model, in conjunction with the Chi-Square Forward Sequential Feature Selection technique, represents a robust and effective solution for precise water quality estimation. This proposed contribution enhances water resource management and environmental safeguarding with an accuracy of 96.82%, offering versatile applications in public health and environmental engineering.