Utilizing Data Mining Algorithms to Predict Groundwater Levels: An Applied Study on the Rising Groundwater Issue in Zliten, Libya

Among the most valuable environmental challenges facing the city of Zliten, Libya, which is depending totally on groundwater resources, is the abnormal rise in groundwater level. The study will try to use data mining techniques for extracting an accurate predictive model for groundwater levels based on monthly observations from 31 monitored wells collected over a continuous period of ten months. In this context, two machine learning-based algorithms for prediction were applied: Linear Regression and Decision Tree Regression, using the specialized analytic environment RapidMiner. These results show that, based on the key statistical performance measures of RMSE, MAE, and R², the Decision Tree algorithm obviously outperformed the Linear Regression model and thus can be effective in forecasting future groundwater levels. Such findings bring into focus the need to adopt artificial intelligence tools to support the management of groundwater resources and environmental sustainability in the face of fluctuating and unstable water supplies.