The scientific and accurate mapping of debris flow susceptibility and the determination of main control factors and their contribution rates are important foundations for regional debris flow warning and risk management. The article takes the upper reaches of the Minjiang River as the research area and small watersheds as evaluation units. Five machine learning models were used to construct an evaluation model for the susceptibility of debris flows in the upper reaches of the Minjiang River, and quantitative analysis was conducted on the susceptibility of debris flows and the contribution rate of evaluation factors before and after the Wenchuan earthquake. The results show that: (1) The ACC
values of the integrated machine learning model are higher than those of the shallow machine learning model, and the random forest model performs best in the assessment of debris flow vulnerability before and after the earthquake; (2) The occurrence rate of debris flow before and after earthquakes gradually increases with the increase of susceptibility level, and the increment increases with the increase of the level. The occurrence rate of debris flow after each level of earthquake is higher than before the earthquake; (3) The contribution rate of the erosion transfer coefficient before and after the earthquake is significantly higher than other factors, which is superimposed with the spatial distribution characteristics of the Wenchuan earthquake intensity, exacerbating the spatial distribution pattern of the gradually decreasing development degree of debris flows in the main and tributaries from downstream to upstream after the earthquake.