In this paper, a decision tree binary regression tree is set up based on the Light GBM model, the region of the tree is divided and the number of leaf nodes of the tree is iterated continuously. The regularization term controls the leaf node splitting and reduces the overfitting of the model to obtain the objective function of the model, the training data set is recorded as the loss function, and the objective function is optimized to optimize the loss function. The features with the maximum information gain are selected as the split nodes for information gain, and the model prediction accuracy evaluation index is determined to obtain the final model. The results show that the highest efficiency can reach 95% after the administrative management efficiency is improved, and the administrative management efficiency can be improved by strengthening administrative supervision in the new era.