Who buys food products from online influencers? predictions with machine learning

Abstract The burgeoning growth of rural e-commerce in China is poised to fundamentally transform the rural economic and social landscape. Within this evolving ecosystem, agricultural online influencers have emerged as pivotal actors, yet their specific roles and underlying mechanisms warrant further investigation. Based on online survey data, this study utilized four machine learning techniques to predict consumer behavior in purchasing agricultural products recommended by online influencers and identified key predictors. We found that the random forest algorithm incorporating a comprehensive set of economic and social predictors, achieves a best prediction accuracy of 85.12%. The random forest algorithm and LASSO regression together identified the key predictors. Notably, purchase intention and whether to follow the online influencer on social media emerged as the most significant features enhancing prediction accuracy. Furthermore, our analysis highlights the importance of demographic factors (age), perceived value assessments (emotional and social value), engagement levels (attention to online influencers), and prior e-commerce experiences as crucial mechanisms. Our findings are of practical values for online influencers and related enterprises to improve marketing strategies, promoting branding and commercialization, and for the development of rural e-commerce in China and related professionals’ cultivation.