An examination of critical factors influencing consumers' behavior in live streaming e-commerce

The rapid development of live streaming e-commerce calls for a deeper understanding of the factors influencing consumer behavior. This paper uses web crawling to gather data, combining Latent Dirichlet Allocation (LDA) and word frequency analysis for feature extraction. Hierarchical regression is applied to validate critical factors influencing consumer behavior in live streaming e-commerce. The data covers March to September 2023, including 361.68 hours of live streaming, 3,093,402 words of anchor speeches, and 4,095,309 records of consumer behavior. Eleven critical factors under six dimensions from three perspectives were identified and the combination of three perspectives has a role in influencing consumer behavior. All of critical factors influenced consumer behavior, with the most significant impacts of the anchor's positivity and promotion, the product's delivery service, and the number of bullet screens in the scene. This paper highlights the combined influence of anchor, product, and scene perspectives on consumer behavior, providing valuable insights for optimization.