Market demand forecast and dynamic optimization of intangible cultural heritage creative products driven by reinforcement learning

The commercialization of Intangible Cultural Heritage (ICH) creative products plays a crucial role in preserving cultural identity while supporting economic development. Accurate market demand forecasting remains challenging due to some issues. Conventional forecasting and optimization methods rely on static models, which fail to adapt effectively to dynamic and uncertain market environments. This gap highlights the need for advanced intelligent decision-making frameworks to enhance the sustainable management of ICH creative industries. Research proposes an Intelligent Bald Eagle Search Deep Q-Network (Int-BES-DQN) framework; Int-BES is used to efficiently explore the global solution space to identify optimal product demand strategies, while DQN leverages adaptive reinforcement learning to continuously improve decision-making based on feedback. The creative products demand dataset contains 3500 records, including historical sales records, product attributes, market competition, consumer behavior profiles, and cultural event calendars collected from online platforms and local marketplaces. Data preprocessing is conducted using Z-score normalization and handling missing-value imputation to ensure data quality and uniformity. Principal Component Analysis (PCA) is used to discover important demand drivers and reduce dimensionality in feature extraction. The proposed Int-BES-DQN aims to enhance market demand forecasting and dynamic optimization by combining BES’s global search efficiency with DQN’s adaptive reinforcement learning capability. The model demonstrates high forecasting (0.92) accuracy, (0.9889) precision, (0.945) recall, and (0.938) F1-score, with reduced (0.02) MSE compared to baseline models. Implemented in Python using TensorFlow and Scikit-learn, this research introduces a scalable reinforcement learning approach linking cultural preservation with market competitiveness.