With the increasing intensity of market competition, businesses are continuously demanding optimization of their marketing strategies. Traditional methods for evaluating marketing strategies face challenges such as difficulty in real-time adjustments and an inability to accurately simulate complex customer behaviors. We proposes a framework for evaluating and optimizing marketing strategy effectiveness based on deep generative models-Gen-Opt. This framework combines Generative Adversarial Networks (GANs), Variational Autoencoders (VAE), and the Autoformer time series prediction model to effectively capture customer behavior patterns, forecast sales trends, and simulate the potential response to marketing strategies. Through experiments on two publicly available datasets, Rossmann Store Sales and Retail Sales Forecasting, the Gen-Opt model outperforms traditional methods and existing mainstream models on various performance metrics. Specifically, on the Rossmann Store Sales dataset, Gen-Opt achieves MSE and RMSE values of 0.345 and 0.587, respectively, while on the Retail Sales Forecasting dataset, the MSE and RMSE are 0.278 and 0.528, showing improvements of approximately 15%-30% over existing methods. The ablation experiment results show that GANs, VAE, and Autoformer modules effectively improve the model's prediction accuracy and generalization ability, with the removal of any one of these modules significantly degrading model performance. Overall, the Gen-Opt model presented in the article provides a new approach for optimizing marketing strategies, effectively addressing the limitations of traditional methods and providing more precise and dynamic support for business decision-making.
Xiao et al. (Tue,) studied this question.