PREDICTING CONCENTRATION CHANGE OF SOME TMS IN SOIL–WATER ECOSYSTEM USING MACHINE LEARNING

Abstract

This article provides a discussion of the applicability of machine learning methods, with a particular focus on linear regression, to predict the total concentration of TMs in the soil–water ecosystem. Despite the requirement of only minute quantities, TMs have the potential to exert a detrimental effect on the environment. For prediction, seasonal and geographical parameters along with metal concentrations in the soil and their irrigation water were used. A key focus of the study was the normalization of data, a process that has been shown to improve the identification of linear relationships between variables. The developed linear regression model demonstrated a high degree of precision as evidenced by the coefficient of determination 0.9945, the average absolute error of 0.1, and the average percentage error of 5.5%. These findings substantiate the feasibility of employing the proposed methodology to monitor water quality, evaluate pollution risks, and identify potential threats at an early stage in ecosystems that anthropogenic factors have been impacted.