<p>Landslides often pose a huge threat to the lives and property of people. The detailed analysis of landslide activity characteristics and landslide hazard assessment can help to understand the activity laws of landslides, and identify the possibility of landslide occurrence, which play a key role to take necessary disaster prevention and mitigation measures. In this research, we selected Eurasia as the study area. Firstly, a landslide dataset of the Eurasian continent was analyzed. Based on the Emerging Hot Spot Analysis, obvious hot spots of landslide hazards around the Himalayan Mountains and new cold spots in the Loess Plateau were found. On the other hand, for large scale regions, there are significant differences in landslide-developing environments, and the effects of various causing factors are different significantly. In this research, the weights of landslide-causing factors for different kinds of mountain-hazards developing environments were obtained through the machine learning method (Random Forest). Different landslide hazard assessment models under different mountain-hazards developing Environments were constructed. It solves the problem of using only one model to evaluate landslide hazard in complex disaster environments at a large regional scale without considering regional differences. This also provides a new idea for evaluating the hazard assessment of other mountain hazards at a large regional scale. The results show that the landslide hazard assessment model has obvious regional characteristics and can better realize the purpose of hazard assessment. About 16.35% of the mountain areas in the Eurasian continent belong to very high-hazard and high-hazard areas of landslides. At the same time, the contribution values of different factors in the calculation process based on the hazard assessment models can better identify the main controlling factors and propose the adoption of more targeted disaster prevention and mitigation measures.</p>