Abstract: Due to the differences in geographical and human conditions among different regions, the importance of evaluation factors for landslide susceptibility also varies dramatically. Therefore, selecting the appropriate evaluation factors is an important prerequisite for revealing the spatial distribution characteristics of landslide-prone areas. By considering the optimal discretization of factors and combining Geodetector（GD） and Spatial Principal Component Analysis（SPCA）, a landslide susceptibility index is constructed. It is found that evaluation factors generally exhibit a trend where the higher the number of discrete categories, the higher the explanatory power of the factors, and both the natural breakpoint method and quantile classification method yield good results. The interaction between factors affects the explanatory power of factors for landslides, with the Normalized Difference Vegetation Index（NDVI） having an explanatory power of up to 0.453, but its coupled explanatory power is much lower than that of population density. The landslide susceptibility index of the study area is 0.589, indicating a moderate susceptibility area, with highly susceptible and extremely susceptible areas mainly distributed in the northern mountainous areas with dense vegetation cover. The landslide susceptibility index of the study area exhibits significant spatial clustering characteristics, with high-high clustering areas mainly distributed in the middle and low mountainous areas in the north of the study area, and low-low clustering areas mainly distributed in the gentle valleys in the south of the study area. This study could provide reference for the landslide susceptibility zoning in Southwest Fujian Province.