Video Quality Assessment Based on Statistical Selection Approach for QoE Factors Dependency

Quality of Experience (QoE) becomes a topic of utmost eminence for service providers and the major factor in the success of multimedia services. Thus, it is challenging to investigate thoroughly the human side of QoE in order to find out the impact of factors that affect user satisfaction. In this paper, we provide a structured way to build an accurate and objective QoE model. In order to serve this purpose, Principal Component Analysis (PCA) and Analytic Hierarchy Process (AHP) approaches are combined and used to select the factors which have a significant impact on user satisfaction and essential for predicting QoE. Random Forest technique is used as a machine learning method to classify original datasets based on real environment, collected in the form of subjective scores. The results show an efficient estimation of QoE with respect to the five most influencing factors (frame rate, video size, audio rate, resolution and mean bit rate).