In this paper, a talent training model based on big data analysis is designed for the background of the construction of the Hainan Free Trade Port. A learning behavior analysis method is designed using the K-Means clustering algorithm and particle swarm optimization algorithm, which can accurately mine valuable information from a large amount of user learning behavior data and provide a reference for the exploration of the Russian talent training model. The accuracy rate in the experimental validation can reach 91.99%, which is outstanding and can provide important support for establishing a systematic talent training model in the context of the Hainan Free Trade Port.