Factors Influencing the Citizenization of Rural-Urban Migrants:A Study Based on Machine Learning
Based on the China Migrants Dynamic Survey(CMDS)data in 2017,this paper systematically investigates the influencing factors of the citizenization of rural-urban migrants using the multiple linear regression,penalized regression,ensemble learning,deep learning,and other machine learning methods.The results show that the ensemble learning method outperforms the multiple linear regression method in predicting the level of citizenization of rural-urban migrants,among which the gradient boosting regression tree(GBRT)model performs the best prediction.Moreover,among all the characteristic variables,the individual's education level,gender,family size,age,and number of mobile cities are the most important influencing factors for the citizenization of rural-urban migrants.In addition,we use the accumulated local effects(ALE)plot to show the specific prediction patterns of different influencing factors,and find that factors such as age and number of mobile cities have non-linear characteristics in their influence on the citizenization of rural-urban migrants.The conclusions of this paper are informative for the design of policies aimed at accelerating the citizenization of rural-urban migrants.
NETL
NSTL
万方数据
