In order to solve the problem that the influence of public satisfaction on the evaluation effect is less considered in the evaluation of the implementation effect of expressway differentiated toll collection. Based on the satisfaction questionnaire of differentiated toll road sections in Yunnan Province in 2022, a gradient boosting iterative decision tree model (GBDT) under the binary classification experiment was constructed to analyze the influencing factors of the satisfaction of differentiated toll policies on expressways. The results show that the gradient boosting iterative decision tree model (GBDT) under the binary classification experiment has good fitting. At the same time, the satisfaction degree of timely and accurate preferential policies is the most important factor affecting the overall satisfaction of expressway differentiated charging policies, and its contribution degree is 20.35%. In addition, the independent effect analysis shows that the overall satisfaction of the differentiated charging method for empty trucks is the highest, and the overall satisfaction of the differentiated charging method for road sections is the lowest.