Refactoring is a crucial software engineering practice aimed at improving code quality. However, detecting and predicting refactoring activities automatically remains a challenging task due to the limited availability of labeled datasets. This study investigates the role of data augmentation techniques in enhancing refactoring detection models. We apply various augmentation strategies to expand training data and assess their impact on model performance. We also discuss validation and balancing of the resulting dataset in order to provide meaningful data for further applied ML techniques. Our findings highlight the importance of data diversity in automated refactoring detection and provide insights into optimizing augmentation strategies for software engineering applications. Experimental results demonstrate that data augmentation improves the robustness and accuracy of refactoring detection models by mitigating overfitting and enhancing generalization.