Determining investment allocation strategies to improve consumer satisfaction based on a preference learning model

Abstract

Mining product attribute performance, importance, and their (a)symmetric impacts on consumer satisfaction from online reviews is crucial for enterprises to formulate real-time investment allocation strategies for product improvement. While existing studies have employed machine learning, regression, and correlation analysis to explore these complex relationships, they face the challenge of balancing prediction accuracy with interpretability. This paper proposes an asymmetric importance-performance analysis model based on preference learning with online reviews. It devises an asymmetric value function incorporating unknown preference parameters to elucidate (a)symmetric impacts of attribute performance on overall consumer satisfaction. The process of learning preference parameters is implemented by mathematical programming with a simulation experiment. Attributes are classified into eight categories according to their performance and importance, each corresponding to an improvement strategy. An optimization model is constructed to develop investment allocation strategies for attribute improvement, aiming at maximizing consumer satisfaction within established financial constraints. A hotel-focused case study showcases the approach, and simulations validate the robustness of the proposed model.