Machine Learning-Based Water Quality Prediction with Grid Search Hyperparameter Optimization

Water is essential for life, but factors like industrialization, contamination, and natural disasters often compromise its safety. The Water Quality Index also known as WQI is commonly used to evaluate the drinking quality of water based on standards set by WHO. However, traditional water quality analysis remains challenging. This study proposes a machine learning (ML)-based water quality monitoring (WQM) system to classify water suitability for drinking. It employs Random Forest (RF) models on a dataset of 11000 instances with seven key features: pH, Sodium (Na), Total Dissolved Solids (TDS), Chloride (Cl), Nitrate Nitrogen (NO3-N), Dissolve Oxygen (DO), and Bicarbonate (HCO3). Data preprocessing includes the techniques like mean imputation and normalization. Various ML models has been evaluated on parameters like accuracy, precision, recall and F1-score. RF model stood out among all models with accuracy of 98.7%. The results highlight the effectiveness of ML-based systems in reliable drinking water quality assessment.