Mean and Prediction Imputation-Based Approach for Predicting Water Potability Using Machine Learning

Water is one of the most critical components of our universe, as it is what keeps us alive. Therefore, the portability of a person's water supply is essential to their survival. Using the water potability indicator, you can determine whether or not the water in question is safe for human consumption as drinking water. Water's ability to support human consumption is inversely related to its concentration of minerals, particularly pH, sulfate, and chloramines, as is well-known; in this study, they took a water portability dataset from Kaggle publicly available. Unfortunately, every dataset includes some null values. Therefore, two approaches are proposed in this study: mean imputation and predicting imputation for replacing the null values. Six machine learning algorithms with each approach are implemented, achieving an accuracy of 66.23% and 70.09%. In the future, the Prediction imputation approach can be implemented on a different dataset.