Climate change is one of the major environmental challenges currently facing the world, and accurate prediction of climate change trends is of great significance for addressing climate change. This study combines machine learning and mathematical modeling methods to construct a comprehensive climate change prediction model. A variety of regression models were used, including traditional time series analysis and advanced machine learning models. Climate data such as global temperature, precipitation, and CO₂ concentration were processed and analyzed to compare the prediction accuracy and stability of different models. The experimental results show that the Long Short-Term Memory (LSTM) model has a clear advantage in long-term trend prediction, especially in capturing the long-term dependencies and complex nonlinear characteristics of climate data. Although machine learning methods have demonstrated good predictive performance, they still need to be combined with physical climate models to further improve the interpretability and reliability of the models. The research results provide more scientific support for coping strategies for global climate change and promote the precise formulation of climate policies.