QoE is almost the most important parameter, estimating which it is possible to determine user experience and compare it with users' expectations. Respective method based on the principles of machine learning has been developed to assess and optimize the state of the network in order to improve the quality of service provision to users by cellular operators using the model of service quality control system. Proposed method is a very effective instrument for request formation from users, network status monitoring, service providing quality control integral marks modelling and their optimization, search of degradation reasons and reporting ranging.