Abstract: Landslides, as one of the most destructive geological hazards worldwide, necessitate risk assessment for ensuring the stability of socio-ecological systems in national parks. Susceptibility assessments are beneficial for the scientific management of landslide disasters and geological disaster risk within national parks. With the Chengdu Panda National Park as the study area, this study selects 20 disaster-causing factors, such as topography, hydrology, and soil vegetation. A hybrid model is established based on the VIKOR method, frequency ratio (FR) method, and a combination of three classification algorithms-Random Forest (RF), CatBoost (CB), and LightGBM (LG). The frequency ratio is utilized to examine spatial correlation between factors and landslide occurrence, while the three algorithms are used to calculate factor importance. Based on these findings, the hybrid model is employed to draft the susceptibility map for landslides in the region. The accuracy and precision of the models are evaluated using the Receiver Operating Characteristic (ROC) curve, Area Under the Curve (AUC), and Kappa coefficient. The results indicate that the VIKOR-FR-LG model demonstrates superior performance in predicting landslide susceptibility. Elevation, proximity to rivers, and distance from fault lines are significant factors influencing landslide occurrence. The area with extremely high sensitivity amounts to 2473.30km². Therefore, it can be inferred that the VIKOR-FR-LG model is an effective tool for predicting landslide susceptibility in the research area. The results of this study can serve as a reference for landslide prevention and ecological safety management in the Chengdu Panda National Park.