The management of international students has become a new challenge that China and most countries in the world must face in the context of the “Belt and Road Initiative” (BRI) and the globalization of higher education. In this paper, we conducted an empirical study on international students in China (ISC) based on a combination of spatial econometric models and analytical methods such as BCG, GeoDetector, and DDCAM, trying to provide a basis for “evidence-based decision-making” by the government in the management of international students and talents. Quantitative analyses revealed significant diversity and spatial effects in the spatio-temporal dynamics of ISCs, with the emergence of star, gazelle, cow, and dog types, suggesting that the scientific management of ISCs requires both local adaptation (spatial differentiation and heterogeneity) and regional collaboration (spatial correlation and autocorrelation). There were many factors affecting the spatio-temporal dynamics of ISCs, and the force of different factors varied considerably, with the key factor more than 0.5 but the auxiliary factor being less than 0.2. Moreover, the factors had significant interaction effects, and a large number of super-interacting factor pairs emerged, with a joint force of more than 0.9, twice the force of a single factor. Generally, the factors had a complex driving mechanism, suggesting that ISC management requires identifying the key and important factors, while attaching emphasis to the synergistic effects among different factors. The recommendation is that China should manage ISCs in spatial zones and introduce differentiated development strategies and policies in different zones. In conclusion, this paper proposes a technical route integrating “dynamic analysis-driving mechanism-policy design” for international students, which is applicable to China, while providing a reference for the management and spatial planning design of international students in other countries in the world.