Abstract: Soil-rock mixture is a geological material with complex physical and mechanical properties. The stability evaluation of soil-rock mixture slope is very important in the field of engineering geology. In order to improve the ability of slope stability prediction, this paper presents a fusion algorithm by combining the particle swarm optimization swith the fruit fly optimization algorithm. Then based on the fusion algorithm, the paper constracts the machine learning prediction models for slope stability of soil-rock mixture. It uses three evaluation indexes to evaluate the accuracy of the models. Finally, the important analysis of input parameters is carried out using the gradient boosting decision tree model based on fusion algorithm. The results of this paper show the following. (1)Compared with the particle swarm and the fruit fly optimization algorithm, the fusion algorithm can effectively optimize the parameters of the machine learning model to significantly improve the prediction accuracy of the prediction model. (2)The prediction model of the gradient boosting decision tree model based on the fusion algorithm has the highest accuracy up to 93.33%, which is obviously better than the decision tree model and the stacking model with the fusion algorithm. (3)The structural factors that influence the stability of the soil-rock mixture slope are, from high to low, the inclination angle of the bedrock surface, the stone content, the overall slope angle, and the slope height.