Evaluation of Sustainable Use of Water Resources Based on Random Forest: A Case Study in the Lishui River Basin, Central China

The evaluation of the sustainable use of water resources is an important part in the study of the ideal allocation of water resources. Selecting relevant evaluation indices with diverse scientific characteristics directly affects the accuracy of the results of water resource assessment. A machine learning method, random forest (RF), was applied to evaluate the sustainable use of water resources in the Lishui River basin of Human province. Python software was used to analyze eight correlating indices of the data set, create the visual matrix diagram of the original data collection and distribution points, and then analyze the evaluation results of this method. Based on a comparison with artificial neural networks (ANN) and support vector machines (SVM), the RF method evaluation results were considered reasonable and reliable with greater stability, accuracy, and practicability, and a classification evaluation standard rate of 99.6%.