Network video traffic constitutes a significant portion of overall internet traffic and continues to grow rapidly, primarily due to the popularity of platforms such as YouTube and Netflix. As data transmission volumes increase, networks face greater challenges in maintaining Quality of Service (QoS), as video traffic demands substantial resources. This results in increased strain on communication channels, higher latency, and reduced service quality. With the prevalence of encrypted traffic, traditional classification methods, such as Deep Packet Inspection (DPI) and port-based classification, face limitations. DPI requires significant computational resources and is ineffective for encrypted traffic, while port-based classification is challenging due to the use of dynamic ports and encryption protocols. This work aims to address these issues through machine learning methods that analyze statistical traffic features and provide high accuracy in video traffic classification. This approach reduces dependency on packet content and enables efficient processing of encrypted content, optimizing the use of network resources.