Wind power generation is an important part of the field of renewable energy, which is of great significance for optimizing the efficiency of wind power generation and the stability of power supply. The analysis of wind speed data can help better predict wind speed changes, so as to optimize the operation and management of wind power generation. Decision trees are a common machine learning algorithm used for classification and regression tasks. In this paper, CHAID, Random Forest and Gradient Boosting Trees, three common decision tree algorithms, are selected for modeling to predict wind speed under different scenarios. By comparing and analyzing the results, they show that the random forest algorithm can achieve relatively high prediction accuracy in wind speed prediction. Random forest algorithm has high accuracy and strong anti-overfitting ability, which provides a theoretical reference for location selection and wind speed analysis of wind turbines.