Providing a model for predicting and status of aquifer depth changes through tree and clustering algorithms

The need for a model for effective planning and management of water resources, particularly groundwater, is especially critical in light of water scarcity and aquifers. Given the importance of various factors in determining the amount of drop, this study used human and natural factors to predict the amount of aquifer drop in Qazvin. To accomplish this, the K-Means clustering algorithm was used first, followed by the tree algorithms CART, CHAID, C5.0, and QUEST to determine the optimal ratio between different fields. Accuracy values of 0.90, 0.96, 0.94, and 0.92 were obtained for the aforementioned tree algorithms. The values obtained for the CHAID algorithm's sensitivity, transparency, accuracy, precision, false-positive rate, false-negative rate, F-measure, geometric mean, and error rate demonstrate that this algorithm outperforms other algorithms. The amount of water in the irrigation network is the most influential human factor in model production, while the amount of temperature is the most influential natural factor. The proposed model enables more accurate prediction of aquifer changes and can be used by managers and farmers to improve aquifer management.