In recent years, the rapid growth of video streaming services has highlighted the need for accurate prediction of Quality of Experience (QoE) to optimize content delivery and enhance user satisfaction. Predicting QoE is a complex task due to the numerous influencing factors, such as network conditions, device types, and content characteristics, in dynamic and large-scale environments. This paper proposes a novel Mixture of Experts (MoE) model for QoE prediction in video streaming. The model employs Incremental Support Vector Machines (ISVMs) as specialized experts, each trained on distinct subsets of influencing factors (IFs), and uses a 1D Convolutional Neural Network (1D-CNN) as a gating mechanism to dynamically select the most relevant experts. ISVMs are specifically leveraged to address the lack of available datasets and to enable real-time, online QoE prediction. The proposed approach aims to enhance prediction accuracy while minimizing computational complexity. Experimental results demonstrate the effectiveness of the MoE model in predicting QoE for video streaming, outperforming existing methods.