Against the backdrop of the rapid development of the digital economy, existing recommendation algorithms have problems such as insufficient accuracy and insufficient flexibility in capturing user preferences. This paper aims to study the personalized recommendation algorithm based on big data analysis. Through big data technology, user behavior data is comprehensively collected, including click stream data, purchase records, and user reviews, and data cleaning and feature engineering are performed to extract user characteristics and product characteristics. The specific steps include: (1) using collaborative filtering algorithm to calculate the similarity between users and make preliminary recommendations based on the recommendations of neighbors; (2) using RNN for feature learning and inputting the feature vectors of users and products into the model to capture nonlinear relationships; (3) updating the model in real time through feedback mechanism and adjusting the recommendation strategy. The accuracy of the proposed recommendation algorithm in the application of a certain e-commerce platform has increased to 99%. The personalized recommendation algorithm based on big data analysis not only effectively improves the performance of the recommendation system, but also provides important support for business decisions in the digital economy.