Consumer Behavior Prediction using GAI tools

This thesis investigates the application and effectiveness of Generative Artificial Intelligence (GenAI) techniques, specifically Generative Adversarial Networks (GANs), Variational Autoencoders (VAEs), and Transformer architectures, in predicting consumer behavior within e-commerce contexts. Using historical consumer data from an e-commerce platform, this study systematically benchmarks GenAI models against traditional machine learning methods, demonstrating the superior predictive accuracy and robustness of Transformer-based models. Additionally, the research validates the practicality and reliability of synthetic data generated by conditional GANs, highlighting significant implications for data augmentation, privacy, and analytics accessibility for small businesses. Practical recommendations are provided for marketers, emphasizing enhanced customer segmentation, personalized promotional strategies, and optimized inventory management. Ethical considerations regarding fairness and transparency in predictive analytics are also discussed, ensuring responsible AI deployment.