Application of machine learning algorithms in the investigation of groundwater quality parameters over YSR district, India

Authors : Jagadish Kumar MOGARAJU

Pages : 64-72

Doi:10.31127/tuje.1032314

View : 15 | Download : 13

Publication Date : 2023-01-15

Article Type : Research Paper

Abstract :Human life sustained for decades due to the availability of basic needs, and freshwater is one of them. However, groundwater quality is constantly under pressure. This can be attributed to anthropogenic activities not limited to urban areas but to rural zones. Machine learning methods like linear discriminant analysis insert ignore into journalissuearticles values(LDA);, Classification and Regression Trees insert ignore into journalissuearticles values(CART);, k-Nearest Neighbour insert ignore into journalissuearticles values(KNN);, Support Vector Machines insert ignore into journalissuearticles values(SVM); and, Random Forest insert ignore into journalissuearticles values(RF); models were used to analyse groundwater quality variables. The mean accuracy of each classifier was calculated, and the obtained mean accuracies were 77.5% insert ignore into journalissuearticles values(LDA);, 87% insert ignore into journalissuearticles values(CART);, 96% insert ignore into journalissuearticles values(KNN);, 93.5% insert ignore into journalissuearticles values(SVM); and 96% insert ignore into journalissuearticles values(RF);. RF and KNN models were selected as optimal models with higher accuracy. This study made it apparent that machine learning algorithms can estimate and predict water quality variables with significant accuracy. In this study, the observations and variables were compared with the water quality index and drinking water limits provided by the Bureau of Indian Standards. The water quality index for each observation was calculated. If at least four variables have a higher value than prescribed limits, it was assigned a value of 1; if more than four variables reported higher values, it was assigned a value of 2. 
Keywords : Groundwater, Machine learning, Support vector Machines, Classification Trees, Random Forest