North China (NC), a transition zone between semi-arid and semi-humid as well as agriculture and livestock systems, is highly sensitive to climate change. Effective prediction of vegetation coverage changes in such transition zones under the context of climate change is critical for maintaining ecosystem stability and enhancing climate change adaptation capacity. In this study, we construct a model to simulate summer anomalies of vegetation coverage in NC based on machine learning. Considering that the change of vegetation coverage depends on regional climate characteristics, we finely divide NC into six climatic sub-regions based on climate similarity using K-means Cluster and then apply eXtreme Gradient Boosting (XGBoost) to characterize the relationship between meteorological elements and vegetation coverage. The integration of K-means Cluster and XGBoost (XGB) effectively simulates the change in summer vegetation coverage in NC, outperforming the traditional linear model by 14% and the Support Vector Machine (SVM) by 5%. XGB can characterize the nonlinear relationship between meteorological elements and vegetation coverage and identify the factors that contribute significantly to vegetation coverage change in specific climate sub-regions. The results indicate that the water factor plays a pivotal role in the semi-arid zone, while the energy factor controls the vegetation coverage in the semi-humid region. By investigating the effects of meteorological conditions on vegetation coverage in the NC, this study helps agriculturalists understand how climate change affects crops and other vegetation.