The paper is devoted to the approach for static analyzer’s warnings classification using machine learning methods. Static analysis plays a crucial role in software development by identifying potential errors and vulnerabilities in source code. However, the huge amount of warnings generated by static analyzers often overwhelms developers, making it challenging to prioritize and address them effectively. This paper proposes a novel approach to automatically classify static analyzer warnings using machine learning techniques. It showed the result up to 92% accuracy on real-world projects. Our method uses a set of code metrics collected during the analysis process to generate features for training the classifier. In the paper it is explored various strategies for feature selection and model training, considering different types of warnings and analyzers. The effectiveness of the proposed approach in warning classification was experimentally demonstrated. The developed method will be used to automatically review the results generated by the industrial static analyzer.