Abstract: Machine learning is widely used in regional landslide susceptibility evaluation. Traditional studies usually use GIS random generation methods to obtain safety points, but such methods have a large arbitrary nature, which makes the machine learning prediction results less accurate and poorly fitted. In addition, the area selected in the previous study was too small, resulting in insignificant intra-regional variation of rainfall, which led to insensitivity of training and invisible neglect of rainfall-causing factors. In view of this, we select the 34 000 km2 northeast of Chongqing City as the study area and introduce an entropy index (IOE)-random forest (RF) coupled evaluation model. The model forms an entropy weight table by objectively calculating the probability density and entropy weight index of each type of influence factor, and then generates a landslide susceptibility zoning map. We select safety points from the low susceptibility area, ensuring they are non-landslide units, which reduces randomness while improves rationality and accuracy. The selected safety points and 581 landslide points in the region are used as training sets, and together with nine influence factors of slope angle, slope aspect, landform, lithology, distance from roads, distance from river, distance from fault, NDVI, and rainfall are imported into the random forest algorithm, so a landslide susceptibility zoning map is obtained by using the GIS platform, which realizes landslide hazard susceptibility assessment. The results show that the IOE-RF model is about 8% more accurate than the single RF model, which proves that the IOE-RF model can be applied to the evaluation of landslide susceptibility in Northeast of Chongqing City or similar large study areas; the east-west trending arc of the Daba Mountains near the alluvial-fold zone in Northeast of Chongqing City and other special high-risk areas should be focused on in the future prevention and control work. The results can provide an auxiliary decision-making basis for local landslide hazard prevention and management.