Ensuring Clean Water Through Machine Learning Driven Water Quality Prediction

The contamination of water resource is a common concern which affects millions of people round the globe. Ensuring clean drinking water is a crucial task to be addressed with utmost priority. Water quality has to be continuously monitored and effective actions has to be taken to ensure safe drinking water. Different methodologies are adopted worldwide to check the quality of water. This research work focuses on machine learning based water quality prediction models. The work aims at building accurate models in identifying potability of water quickly and effectively. The study employs 3,277 samples from nine cities and over six years data gathered from the Central Pollution Control Board (CPCB) of India. Variables that are quantified include pH, hardness, dissolved solids chloride ammonia, sulphate, conductivity, TOC, TTHMs, turbidity and potability. Different ML such as Decision Tree, Random Forest, Logistic Regression, KNN, and SVM are executed and evaluated in this study. The evaluations shows that the SVM model had the highest accuracy of around 85% followed by KNN with 82% accuracy. The study concludes that these models can be safely and easily adopted in water quality prediction rather than investing in expensive apparatus. The designed model is easy to scale and offers simple solution for water quality prediction with reduced complexity and improved accuracy. The model can be further integrated with IoT devices to retrieve the information in real-time and deep learning strategies can be embedded to take the predictive analysis one step further.