Application of Machine Learning to Predict Water Potability

Abstract

Determining water potability is an important area of research with heavy implications for human health and viability of certain living quarters. In this endeavor, I applied machine learning techniques to predict the potability of water. Based on a data set including chemical and physical water parameters, supervised learning programming is applied to this binary classification problem. Neural network mechanisms are applied to determine the significant parameters that impact potability and accuracy in predicting water potability. Through this research, the significant parameters that affect water potability were identified and used to determine whether a particular combination makes the water potable. Additionally, accuracy of in predicting the potability of the water was found to increase when hidden layers were increased and not a significant impact was realized when increasing the epoch value.