Successful multimedia service deployment in advanced networks like 5G and 6G requires efficient Quality of Experience (QoE) management. This demands efficient tools for tracking, estimating, and managing service quality, with machine learning (ML) acting as a crucial enabler. However, since it depends on so many variables and must adjust to dynamic, massive data environments, predicting QoE in multimedia streams is challenging. The intricate relationships between these variables and QoE may be quantified with the use of ML approaches. In this work, we propose an approach for assessing the QoE of video streaming using an enhanced mixture of classifiers. To improve prediction accuracy, our approach uses ensemble learning models, specifically XGBoost, as experts to improve prediction accuracy and handle imbalanced datasets, which is a common issue in real-world scenarios. We also employ a robust gating network trained to minimize an improved error function that combines classification loss with gating loss. A comprehensive simulation model that covers a range of wireless network settings and video sequences is used to evaluate our model. The experimental results demonstrate significant improvements in the QoE prediction.