Measuring the Quality of Experience (QoE) has become an important research topic, especially in video streaming networks. Since it directly measures customer satisfaction, QoE also becomes an interesting topic for network operators. Objective QoE Assessment outperforms in applicability the high-cost offline subjective procedures. However, quantifying and modeling the QoE is still a task with many open challenges. The P.1203 recommendations by the ITU-T are distinguished as the first standard for modeling the video streaming QoE. This work aims to provide the network operators an early warning about the possible next QoE degradation to allow them to avoid this situation proactively. A low-cost user interaction simulation is proposed using Selenium instead of offline subjective tests. Basing on the ITU-T P.1203 recommendations, the video parameters are also extracted and processed to build a training dataset obtained from a real online environment. In addition, a network prob at the client side is used to obtain additional non-streaming network parameters from different providers’ networks during each viewing session. The QoE then has been predicted using machine learning in three different proposed scenarios using both classification and regression. Feature ranking is also used for dimensionality reduction and enhancing the accuracy and training time. The results are compared to the labeled samples of ITU-T standard using cross-validation accuracy, RMSE, and confusion matrices. The results showed a high prediction accuracy and consistency with the ITU-T standards. Also, they highlighted the main parameters that are highly impacting the end-to-end experienced quality of video streaming.