Data-Driven User Perception Prediction in Broadband Services: A Multidimensional Approach

Given that broadband user perception assessment has traditionally relied on empirical rules and is typically unidimensional, this study develops a multidimensional broadband user perception prediction model leveraging big data modeling technology. By integrating network performance and customer service data, the model enables a quantitative assessment of user perception and pinpointing root causes, thereby providing telecom operators with actionable insights for perception improvement strategies. This study identifies the critical data influencing user perception and proposes an algorithmic framework and methodology for multidimensional, two-level modeling. Through innovative algorithms, extensive sample training, and iterative model refinement, the prediction accuracy is significantly enhanced. The research findings have been applied in a production environment, leading to the establishment of a comprehensive user perception evaluation system for operators. This system facilitates the efficient resolution of issues for users experiencing degraded service perception. The dissemination of these research outcomes offers valuable theoretical and practical references for both the academic community and the telecom industry.