Development of a Regression Model for Predicting Ground Subsidence Risk Using Machine Learning

Ground subsidence occurs primarily in urban areas. The main causes of this damage are underground facilities and excavations. Ground subsidence presents significant risks to life and property, which causes social disruption and emphasizes the need for preventive measures. Recent studies have identified the causes of ground subsidence and developed effective risk management strategies. Because ground subsidence results from the complex interactions of multiple factors, a comprehensive approach to risk management is required. In this study, a machine learning-based regression model was developed to predict the risk of ground subsidence. The dataset included historical subsidence records, underground utility attributes, permeability coefficients, soil layer thicknesses, and elevation data for the study area. The collected data were preprocessed using the ArcGIS software suite. Additionally, condition-based datasets were constructed. These datasets were used to train the machine learning algorithms, and the performances of the models were evaluated using the root mean square error (RMSE) and coefficient of determination (R2) to enable the selection of the best-performing model.