基于空间交互作用的中国省际人口流动模型研究

Study on the Model of Inter-provincial Population Flow in China Based on Spatial Interaction

省际人口流动是人口在地点之间转移而形成的嵌入在空间中的有向流网络。理解和预测人口流动网络中流量模式的形成机制及连边的交互强度不仅可以及时揭示新形势下人口流动空间格局分布特征,而且对于合理制定人口发展战略具有重要意义。鉴于此,在统一模型(Universal Model,UM)的基础上,同时考虑空间交互强度在距离和就业数量上的衰减机制,提出了改进的统一模型(Improved Universal Model,IUM)。基于我国2020年全国人口普查的相关数据,从空间结构特征及模型预测误差两个维度,对人口权重机会模型(Population-Weighted Opportunity,PWO)、UM和IUM的预测网络与真实网络进行对比分析,给出三种模型的预测精度结果。结果表明:从空间结构上看,就整体分析可知,三个模型的预测结果均符合真实情况,即我国人口流动的空间格局呈现出“多核心—网格化分布”的结构特点,局部聚集效应明显,呈现“小世界”特征;就个体分析可知,UM与IUM的出入强度分布、相对流出力及相对吸引力预测更符合真实情况,除个别省份模型模拟的结果与真实结果出入较大外,整体上模型模拟的结果较为良好。从误差估算上看,IUM的总误差率最低,且IUM的空间交互效应的相对误差率较之UM有了很大改善,表明其能更好地量化空间交互作用。最后,通过交叉验证分析,表明IUM的预测结果具有一定的稳健性。因此,IUM可以对我国省际人口流动的空间交互强度进行精准预测,具有较好的应用价值。

Inter-provincial population flow is a directed flow network embedded in space formed by the transfer of population between places.Understanding and predicting the formation mechanism of flow patterns and the interaction intensity of connected edges in the population flow networks can not only reveal the distribution characteristics of the spatial pattern of population flow under the new situation in a timely manner,but also has important significance for the rational formulation of population development strategies.In view of this,this paper proposes an Improved Universal Model(IUM)based on the Universal Model(UM),considering the attenuation mechanism of spatial interaction intensity in distance and employment quantity.Based on the relevant data of China’s 2020 population census,from two dimensions of spatial structure characteristics and model prediction errors,the prediction networks of Population-Weighted Opportunity(PWO),UM and IUM are compared with the real network,and the prediction accuracy results of the three models are given.The results show that:In terms of spatial structure,the overall analysis shows that the prediction results of the three models are in line with the real situation.The spatial pattern of population flow in China shows the structural characteristics of“multi-core grid distribution”,with obvious local aggregation effect and“small world”characteristics.As far as individual analysis is concerned,the intensity distribution,relative flow force and relative attraction prediction of UM and IUM are more consistent with the real situation.The results of model simulation in some provinces are quite different from the real results,and the results of model simulation are relatively good on the whole.From the perspective of error estimation,the total error rate of IUM is the lowest,where the relative error rate of spatial interaction effect of IUM is much improved compared with UM,indicating that it can better quantify the spatial interaction.Finally,through cross-validation analysis,the results show that the prediction results of IUM have certain robustness.Therefore,IUM can accurately predict the spatial interaction intensity of inter-provincial population movement in China,and has good application value.