 This study found that the CNN model performed the best among the three algorithms tested, while the LR model performed the worst. Additionally, the CNN model produced a more accurate prediction of the distribution of susceptible areas than the other two algorithms. Furthermore, the accuracy of LSM can be improved with the advancement of algorithms, such as selecting more representative features, which serves as a more effective guide for land use planning in the study area, or other highlands where landslides are frequent. This article was authored by Zhu Liang, Wei Pingpeng, Wei Lu, and others.