<p>Support vector machine (SVM) is a widely used binary linear classification model in supervised learning. This study introduces a globalized distributionally robust chance-constrained (GDRC) SVM model based on core sets to address data uncertainties and provide a robust classifier. The globalization emphasizes uncertainty within the sample population as a whole, rather than focusing on small perturbations around each sample point. The uncertainty is mainly characterized by confidence regions of the first- and second-order moments. Core sets are constructed to capture some small regions near the potential classification hyperplane, improving the classification quality through the expected distance constraint of the random vector to these core sets. Under appropriate assumptions, we derive an equivalent semi-definite programming reformulation of the proposed GDRC SVM model. To address large-scale applications, an approximation approach based on principal component analysis is incorporated. Numerical experiments are presented to illustrate the effectiveness and advantages of the proposed model.</p>