Research on Deep Learning-based Model for Consumer Behaviour Prediction and Business Decision Optimization

In today’s digital economy, companies are facing increasing challenges to accurately predict consumer behaviour and optimise marketing strategies in a huge and complex data stream. This study proposes a deep learning framework that incorporates Long Short-Term Memory(LSTM) and attention mechanisms for accurate consumer behaviour prediction and business decision optimization. A three-level consumer profiling system is innovatively constructed and non-linear dimensionality reduction of features is achieved through self-encoders. A deep Q-network for optimising marketing strategies was designed to model the marketing environment as a Markov decision-making process. Experiments validate the good performance of the model in various business scenarios, especially in capturing consumers’ long-term behavioural patterns. The application results show that the model can effectively improve the marketing efficiency, customer satisfaction and brand loyalty of enterprises, and provides new ideas for business intelligence decision-making in the digital economy.