带自适应Lasso罚的科技人才流动多水平模型研究

随着高水平科技人才竞争日益激烈，科技人才流动受到全球政治、经济、文化和地理等诸多因素的影响。对于不同的经济体，筛选不同水平下影响科技人才流动的关键因素，并估计出这些因素的影响程度，成为颇具价值的研究课题。在真实数据中，由于变量数量众多、数据存在缺失、不同经济体影响因素差异较大，现有模型面临诸多挑战。本文通过利用带自适应Lasso罚的分位数回归，构建缺失数据插补后的科技人才流动模型，以刻画不同分位数水平下全球各经济体科技人才流动的重要影响因素，以期为政府及相关部门提供决策参考。

S&T brain drain multi-level model research with adaptive Lasso penalty

With the increasingly fierce competition, brain drain is affected by many factors such as global politics, economy, culture and geography. For different economies, screening the key factors affecting the brain drain at different levels and estimating their degree of influences have become a valuable research topic. In real data analysis, because of the large number of variables, missing data, and the large differences in influencing factors of different economies, the existing models face many challenges. This paper uses quantile regression with adaptive Lasso penalty to construct a global brain drain model after imputing missing data, so as to portray the important factors affecting the brain drain in various economies around the world at different levels, in order to provide advice in decision-making for the government and related departments.