In the actual scenario of current marketing strategy decision-making, enterprises often face the dilemma of accurately balancing multiple marketing objectives while effectively responding to market dynamic changes. To address this issue, this paper innovatively constructs a multi-objective programming (MOP) model supported by long short-term memory networks (LSTM). When solving the model, we used particle swarm optimization algorithm, and through the operation of this algorithm, we finally obtained the optimal strategy combination that can consider all marketing objectives. The empirical research results indicate that the proposed MOP-LSTM method achieves higher marketing profitability compared to the traditional Multi-Objective Programming (MOP) baseline, yielding an average relative increase of 14% (from 19.3% to 22.0% in average rate of return over the test period). The MOP-LSTM model demonstrates better adaptability, with the average absolute percentage error (MAPE) of sales volume prediction within the test period controlled within ±6%. It provides a more scientific and accurate planning solution for marketing strategy formulation in complex and changing market environments, effectively promoting the achievement of various marketing goals.
Wu et al. (Tue,) studied this question.