Application of autonomous intelligent customer behavior prediction model based on deep learning in retail marketing strategy optimization

Abstract

This study aims to develop and evaluate a deep learning-based autonomous intelligent system for customer behavior prediction and marketing strategy optimization in the retail sector. A hybrid architecture combining Long Short-Term Memory (LSTM) networks with Transformer models in a multi-task learning framework was designed. Evaluation included offline cross-validation and online A/B testing using 1.5 million customer interactions, followed by a 12-month case study implementation in a multinational e-commerce platform. The model achieved a 15% increase in AUC-ROC for purchase prediction and a 22% improvement in Mean Average Precision for product recommendations compared to state-of-the-art benchmarks. The case study revealed substantial enhancements in click-through rates (35%), conversion rates (28%), and customer retention (22%). The hybrid LSTM-Transformer model with a multi-task learning framework significantly outperforms traditional methods, demonstrating the effectiveness of deep learning for customer behavior prediction and marketing optimization. Retailers can leverage this system to enhance personalized recommendations, optimize pricing strategies, and improve customer engagement, resulting in measurable business performance improvements across diverse retail segments.