The synthesis of rail transit stations is measured to reflect the status of urban land use and the vitality of urban space. The harmonious development between rail transit stations and urban land is found to have a significant positive impact on optimizing urban traffic and improving land use efficiency. Information entropy was first employed to measure the functional diversity of the site, and further, dominance and evenness were utilized to explore the spatial distribution and structural characteristics of the functional evolution of the site. Finally, the XGBoost-SHAP machine learning model was adopted to discuss the driving factors and their interaction effects. The results show as follows. From the perspective of the time dimension, the functional information entropy of the site exhibits a generally fluctuating downward trend from 2012 to 2023. From the perspective of the spatial dimension, it is observed that the majority of sites within the Second Ring Road are characterized as changing, while sites between the Second Ring Road and the Fifth Ring Road remain relatively stable. Significant changes are noted in sites located from the Fifth Ring Road to the Sixth Ring Road, resulting in overall instability in site functionality dynamics. The proportion of changing sites is reported to be as high as 76%. From the perspective of structure, significant changes are observed in the functional composition of sites, marked by the growth of commercial service and transportation sites, and the decrease or fluctuation of residential, industrial, and public sites. From the perspective of driving factors, the number of POI (point of interest), road network density, and gross product within the range of stations are identified as the primary factors influencing the functional mixing degree of Beijing Metro stations, with interactive effects among all these primary factors. The findings of this study hold great significance for enhancing the efficiency of land use around subway stations and boosting the vitality of subway stations.