Comprehensive Water Quality Analysis and Prediction Using Ensemble Machine Learning Models

Ensuring water quality is crucial for public health and ecological balance. With increasing urbanization and industrialization, water sources are often contaminated, necessitating regular monitoring and predictive modelling. This study focuses on predicting water quality in Chennai's Puzhal Lake, Retteri Lake, and Chembarambakkam Lake using nine machine learning algorithms: Logistic Regression, Ridge Regression, Stochastic Gradient Descent (SGD) Classifier, Support Vector Classifier (SVC), Nu-Support Vector Classifier (NuSVC), Decision Tree, Random Forest Classifier, AdaBoost Classifier, and XGBoost Classifier. Water quality parameters such as pH, hardness, chloramines, sulphate, conductivity, organic carbon, trihalomethanes, turbidity, potability, Total Suspended Solids (TSS), Total Dissolved Solids (TDS), and Total Solids (TS) were measured using a water test kit. The dataset was used to train and evaluate the machine learning models based onWorld Health Organization (WHO) guidelines. Among the algorithms, the Random Forest Classifier achieved the highest accuracy with a score of 0.8179, followed by Support Vector Classifier (0.7132) and XGBoost Classifier (0.5925). A novel Ensemble Integration and Aggregation Algorithm (EIAA) was proposed to enhance predictive performance by aggregating the predictions of the individual models through majority voting and averaging methods. The EIAA demonstrated improved accuracy and reliability in predicting water quality. This study provides valuable insights into the most effective machine learning techniques for water quality prediction, contributing to improved water management and safety in Chennai.