Landslide Likelihood Prediction using Machine Learning Algorithms

The supply of electricity via power plants is critical to the operation of many critical infrastructure systems in modern society. Natural hazards can disrupt the power supply, cause power outages that can halt economic growth, and impede emergency response until power is restored. The proposed work aims to predict the landslides likelihood in these critical infrastructure locations in the Northeastern USA using integrated databases of explanatory variables and machine learning algorithms. First, data related to landslides are obtained and merged, including topographic, soil moisture, and precipitation-related data. Five regression algorithms, namely: Random Forest, Extreme Gradient Boosting (XGBoost), K-Nearest Neighbor regression (KNN), Linear Support Vector Regressor (SVR), and Linear regression, are utilized to predict the landslide probability and evaluated on the dataset. The accuracy of the models is assessed by using statistical metrics such as mean absolute error (MAE), mean squared error (MSE), and root mean squared error (RMSE). The study results show that Random Forest outperformed other models with the mutual information feature selection method. It achieved an MSE of 0.0011 with mutual information-based feature selection and an MSE of 0.00157 without feature selection. KNN regressor outperformed the other models with an MSE of 0.00139 with correlation-based information selection. The proposed landslide identification model with Random Forest algorithm shows outstanding robustness and great potential in tackling the landslide likelihood prediction by employing ML algorithms.