Drinking water quality estimation of the fountains in bayburt with machine learning methods

Water, often referred to as "blue gold," plays a critical role in influencing the survival, quality of life, and health of living organisms. The presence and quantity of substances within water can have either beneficial or detrimental effects on life. In this regard, the World Health Organization (WHO) has conducted studies on water quality, developing guidelines that specify the permissible substances and their concentrations. In Turkey, similar efforts have been undertaken, with standards established through the 2005 Regulation on Water Intended for Human Consumption. With advancements in technology, artificial intelligence has increasingly been employed in water quality studies, utilizing machine learning techniques to predict water quality. This study focuses on predicting water quality in Bayburt province using machine learning methods, based on microbiological and chemical water analysis data from the city's mains water supply, as well as microbiological water analysis data from public fountains in Bayburt. Initially, the data underwent preprocessing to ensure compatibility with machine learning applications. Subsequently, water quality predictions were performed using various machine learning methods, including logistic regression, k nearest neighbors (knn), support vector machines (svm), naive bayes, decision trees, and random forests. The results indicated that, for the mains water analysis data of Bayburt, the decision tree method was the most suitable. In contrast, for the public fountain water analysis data, naive bayes, random forest, and decision tree methods yielded identical results, suggesting that all three methods are equally viable. The study further concluded that specificity is a crucial metric in water quality prediction studies and should be incorporated into model evaluation processes.