Enterprise retail price prediction method based on improved HPO-LSTM algorithm

Abstract

In modern market economy, accurate price prediction plays a pivotal role in inventory management, marketing strategy formulation, and profit maximization for enterprises. To optimize resource allocation and improve economic efficiency, this research proposes an optimization algorithm that combines Hunter-Prey optimization algorithm and long short-term memory network, while introducing attention mechanism and Q-learning algorithm to optimize the model. In the research results, the improved Hunter-Prey optimization-long short-term memory algorithm showed significantly better prediction accuracy than the control model, with a root mean square error of 0.48 and a mean absolute error of 0.20. At the same time, after the algorithm converged, the accuracy tended to 0.972 and the recall tended to 0.921. In practical business applications, this algorithm could significantly reduce inventory costs (15.2%) and promotion costs (20.3%), while increasing sales revenue (15.4%) and profits (20.4%). The results indicate that the improved Hunter-Prey optimization-long short-term memory algorithm can effectively reduce prediction errors and strengthen the robustness and adaptability of the model. Research can provide enterprises with an efficient and accurate retail price prediction tool, which can help optimize inventory management, reduce resource waste, and enhance market competitiveness.