Based on the data released by UNESCO on international student mobility in tertiary education from 2001 to 2015, this paper draws on network science to construct a multidimensional weighted directed network heterogeneity model. This model combines the GIS spatial analysis method and the negative binomial regression model to study spatiotemporal evolution and factors influencing international student mobility networks. The results are as follows. First, in terms of the evolutionary characteristics of time series, the linkages of international student mobility networks have increased from 4921 to 9137, and its average weighted degree centrality has rapidly increased from 8004 to 20,834, which indicates that both the number of international students and the choices of overseas routines are mounting. Second, in terms of the topological structure, it has evolved from a single core comprising the United States into dual cores comprising the United States and China. Gradually, Asian countries have become the main body of international student mobility networks, while the centrality of European countries has reduced. Thirdly, in terms of spatial structure, the international student mobility network has a significant core-periphery structure and hierarchical characteristics. The countries in the core, strong semi-periphery and semi-periphery alternate with countries from other tiers. The spatial patterns of the international student mobility network "from east to west, and from south to north" are changing. China and Australia have become the new regional centers of international student mobility networks, meaning that regionalization has become an important trend. Finally, regression analysis shows that the flow of international students correlates positively with the quality of higher education in destination countries, the number of students in colleges and universities, the per capita GDP, the historical links connection between the destination country and the country of origin, and the level of globalization. Geographical distance and linguistic proximity have a significant negative correlation with the flow of international students, indicating that the impediment of language is rapidly weakening, while geographical distance still hinders student mobility.

基于联合国教科文组织公布的全球高等学校留学生流动数据,借鉴网络科学,利用社会网络分析方法和负二项回归模型,对2001—2015年全球留学网络时空演化及其影响因素进行研究。主要结论为：① 时序演化特征上,全球留学网络关系流从4921个增加到9137个,平均加权度从8004迅速增加到20834,表明全球留学网络不仅学生数量在增加,且留学路径数量也在不断增加。② 拓扑结构上,从以美国为单一核心,演变成以美国、中国为双核心的拓扑结构,亚洲国家逐渐成为全球留学网络的主体,欧洲国家的中心性逐渐萎缩。③ 空间结构上,全球留学网络具有明显的等级层次性,美国是全球性主导型节点,德国、捷克是区域性主导型节点,次级主导型节点从25个增加到32个,全球留学网络“东向西、南到北”的地理空间格局正在发生变化,中国、澳大利亚等国成为新的区域性中心,区域化成为全球留学网络一个重要趋势。④ 回归分析发现,留学目的地国高等教育质量、高校在校生数量、人均GDP、留学目的地国与留学生来源国间的历史联系及全球化水平与留学生流量呈显著正相关关系,地理距离、语言临近性与留学生流量呈显著负相关关系,表明语言差异的阻抗作用在迅速减弱,地理距离仍然是阻碍学生流动的一个重要因素。

High-quality talents are reshaping the world economic landscape by transcending geographic flows and driving global innovative activities. Based on the data of international student mobility, this paper draws on complex network theory to construct a multidimensional weighted directed network heterogeneity model. This model uses the GIS spatial analysis method to study spatiotemporal evolution of global talent mobility network complexity from 2001 to 2015. The results are as follows. First, the scale of the global talent mobility network is expanding rapidly, and the relationship is becoming closer and closer. In the network, both the number of talents and the choices of overseas routines are mounting. The network development has obvious small-world characteristics. However, the two levels of network differentiation are significant, and the difference is decreasing year by year. Second, the rank-size distributions of network nodes weighted degree of accession and weighted degree of output conform to the law of power distribution, showing a typical "pyramid structure" characteristics, reflecting that the global talent mobility network is controlled by a small number of pivotal node countries. The spatial patterns of the global talent mobility network are basically "from east to west, and from south to north", but the trend of mobile regionalization is gradually emerging, and the role of emerging countries attracting talents has gradually increased. Thirdly, the core-periphery structure of the global talent mobility network is remarkable. The countries in the core, strong semi-periphery and semi-periphery alternate with countries from other tiers. There is a strong mobility of relations between the core countries. The marginal countries are not connected with each other, or have weak links with each other. The talent of the marginal countries mainly flows to the core countries and semi-marginal countries. Finally, the global talent mobility network community has been significantly differentiated. The network has evolved into six associations, including American associations, EU associations, Chinese associations, South American and South African associations, Malaysian associations, and CIS associations. The scale of associations varies widely. Like the overall network, each community has a similar "pyramid structure" feature.

高质量的人才跨越地理的流动影响和驱动着全球范围内的创新活动,正在重塑世界经济格局。研究基于全球高校留学生流动数据,构建加权有向网络模型,对2001—2015年全球人才流动网络复杂性时空演化进行全面刻画。主要结论如下：① 全球人才流动网络规模迅速扩大,网络发育具有明显的小世界性,两级分化显著。② 全球人才流动网络呈典型“金字塔结构”特征,基本呈“东向西,南到北”的地理格局,新兴国家作为人才吸纳国的角色逐渐上升。③ 全球人才流动网络核心-边缘结构显著,核心国家间存在较强的关系流,边缘国家间联系较弱,其人才主要流向了核心国家和半边缘国家。④ 全球人才流动网络分化成美国社团、欧盟社团、中国社团、南美和南非社团、马来西亚社团、独联体社团等6个凝聚子群。

自户籍制度放宽以来,中国经历了大规模的人口迁移。其中,异质性劳动力（高技能劳动力、普通劳动力）在迁移过程中表征出不同的空间格局和网络组织特征,对地区经济发展产生各异的影响。本研究基于人口普查和抽样调查微观抽样数据,使用复杂网络理论对1995—2015年中国异质性劳动力迁移的时空格局和网络演化展开分析,并分析了格局背后的可能成因。研究发现：① 省际高技能和普通劳动力迁移均呈现出持续高度不平衡的空间集聚特征,承载大量人口的迁移流主要由中国中西部地区指向东部沿海地区,但这种空间不平衡特征呈现出一定的减弱趋势。高技能劳动力的平均迁移距离比普通劳动力更长;② 两类劳动力迁移网络中均呈现明显的“小世界”特性,但普通劳动力迁移网络的迁移强度和关联程度均高于高技能劳动力迁移网络;③ 两类劳动力迁移网络均呈现出以北京、上海、广东为主要核心节点的网络结构。高技能劳动力网络结构呈现相对稳定的特征,而普通劳动力网络结构则呈现出变化的趋势;④ 地区经济差异、路径依赖效应、异质性劳动力在就业市场上的差异性以及劳动力市场对异质性劳动力需求的差异是导致两类劳动力迁移格局差异性的重要成因。

With the rapid development of urban economy and the promotion of urbanization, great changes have taken place in the spatial pattern of population migration in China. Migration has played an important role in the economic, social, and cultural development of cities. Based on the provincial population migration matrix data of 31 provinces, autonomous regions, and municipalities in 1985-2015, a social network analysis was conducted to investigate the overall characteristics of interprovincial population migration network and the role of each node. The results show that: (1) Network density of interprovincial population migration has experienced a slow decline in 1985-2000 and a continuous increase in 2000-2015. The concentrated migration of population from the northwest, southwest, and central regions to the eastern developed areas has become the mainstream of development. (2) The inward center potential of population migration network rapidly increased in 1985-2005 and gradually declined in 2005-2015. It indicates that the number of important population moving areas were increasing gradually (such as Tianjin and Fujian), towards a "multi-polarization" development. The outward center potential was always at a low level and exhibited a trend of fluctuation. The inward center potential was clearly larger than the outward center potential. It indicates that the population migration area is relatively concentrated, and the population emigration area is relatively dispersed. (3) The interprovincial population migration network can be divided into 5 groups: Northeast-North China population linkage area, Central Plains-Yangtze River Delta population linkage area, Middle South-Pearl River Delta population linkage area, southwest population linkage area, and northwest population linkage area. Among them, the Central Plains-Yangtze River Delta population linkage area was the most closely linked between and within group; Middle South-Pearl River Delta population linkage area had the highest total population migration. The Northeast-North China population linkage area had a stronger internal connection and only strongly linked with the northwest region externally. (4) According to the population migration analysis, the 31 provinces, autonomous regions, and municipalities can be divided into three major categories: net population migration areas including eight provinces, autonomous regions, municipalities (Guangdong, Zhejiang, Beijing, Jiangsu, Shanghai, Tianjin, Fujian, and Xinjiang); population balanced areas including six provinces and autonomous regions (Hainan, Inner Mongolia, Tibet, Qinghai, Ningxia, and Liaoning), and population net emigration areas including 17 provinces, autonomous regions, and municipalities including Sichuan, Henan, and so on. (5) Based on the centrality analysis and the core-periphery structure, contributions of the provinces, autonomous regions, and municipalities to population migration was very different. In the interprovincial population migration network, a few core nodes had strong control over the total amount and path of population migration. About 25-30% core provinces controlled 70-80% of the population migration and path of the country, and the control power continued to rise. Among these provinces, Guangdong and Beijing had overall control significance. Jiangsu, Shaanxi, Sichuan, and Zhejiang were largely among the core control nodes. The role of Shanghai, Tianjin, Inner Mongolia, and Fujian as core areas had gradually increased. The role of Henan, Hubei, Hebei, and Xinjiang as core areas had declined. The overall migration path of the country is northward.

随着中国城市经济的发展以及城镇化的快速推进,人口迁移的空间格局发生了重大变化。本文基于1985-2015年全国省际人口迁移矩阵数据,运用社会网络分析方法,探讨中国省际人口迁移网络的整体特征以及每个节点在网络中的地位作用。研究结论为：①省际人口迁移网络密度经历了1985-2000年的缓慢下降、2000-2015年持续上升的过程,与西北、西南和中部主要人口迁出地向东部经济发达地区的集中性迁移成为发展主流密切相关。②人口迁移网络的内向中心势明显大于外向中心势,说明人口迁入地相对集中,人口迁出地相对分散。③省际人口迁移网络可分为东北—华北人口联动区、中原—长三角人口联动区、中南—华南人口联动区、西南人口联动区、西北人口联动区5个子群。其中,中原—长三角人口联动区是群内和群间联系最紧密的区域;中南—华南人口联动区是全国人口累计迁移量最高的区域;东北—华北人口联动区内部联系较强,对外仅与西北人口联动区联系较强。④基于中心性分析以及核心—边缘结构划分,各省(市、区)对于人口迁移网络的贡献意义不同,广东、北京具有全局控制意义,江苏、陕西、四川、浙江基本处于核心节点序列,上海、天津、内蒙古、福建核心作用逐渐增强,河南、湖北、河北、新疆核心作用出现下降,全国人口迁移路径总体向北偏移。

在当今全球化与地方化、区域化的背景下,物质和能量在各节点间的高速流动促进了城市网络的形成并成为一种新的区域组织模式和空间结构,尤其是对区域一体化高度发达的区域产生了深刻影响。以长江三角洲地区为例,从人口迁移的网络空间入手,从网络密度、中心势等角度对1982-2010年长三角地区人口网络的演变进行研究。结果表明：整体上人口迁移网络日趋成熟,但空间分布不均衡；人口迁移主要流向上海、杭州、南京、宁波和苏锡常等核心城市,同时这些城市的人口外迁现象逐渐显现；以无锡、苏州、杭州之间人口迁移联系为主体的省际间人口流动行为增多,空间上具有等级扩散的特征；不同空间尺度的网络结构相互嵌套,在地方尺度下形成了江苏以邻域渗透为主和浙江的等级辐合两种网络结构。最后从就业机会、收入水平、产业结构、迁移成本等方面分析了人口迁移网络的演化历经均质离散—单核心集聚—多核心等级网络—链式空间网络四个阶段的成因。

Population migration, social check-in, vehicle navigation, and other spatial behavior big data have become vital carriers characterizing users' spatial behavior. The big data used in this paper were collected from the locations provided by hundreds of millions intelligent mobile phone users through Location Based Service (LBS) Tencent Migration data platform, and were displayed by means of real-time heat map which indicates user’s moving trajectory in China. "Tencent Migration" big data can real-timely, dynamically, completely and systematically record population flow routes using LBS device. Through gathering residents daily mobility among 299 cities in China during the period of "National Day-Mid-Autumn Festival" (NDMAF) vacation (from September 30 to October 8) in 2017 in "Tencent Migration" and defining three periods with "travel period, journey period, return period", this paper is designed to analyze and explore the characteristics and spatial patterns of daily flow mobility cities from the perspective of population daily mobility distribution levels, flow distribution layers network aggregation, spatial patterns and characteristics of the complex structure of the flow network. Results show that "Tencent migration" big data clearly discovers the temporal-spatial pattern of population mobility in China during the period of NDMAF. The net inflow of population showed a diamond-shaped pattern with cross frame support in each period, with the four nodes of Beijing, Shanghai, Guangzhou and Xi’an. Main mobility assembling centers are distributed in the urban agglomerations of Beijing-Tianjin-Hebei, Yangtze River Delta, Pearl River Delta and Chengdu-Chongqing, and those centers have strong coherence with those urban hierarchies. There is a positive correlation between the level of urban administration and the influence of population flow. Most cities are in a state of "relative equilibrium" in the population flow, and clear hierarchical structure and level distinction can be identified. Spatial patterns of population mobility present obvious core-periphery structures. The Dali-Hegang line exhibits a significant network of spatial differences in terms of boundary divisions. In this context, the spatial distribution of urban network could be summarized as "dense in the East and sparse in the West", and the core linkages of urban network could be characterized as "parallel in the East and series in the West". The whole network exhibits a typical "small world" network characteristic, which shows that China's urban population flow network has high connectivity and accessibility during the period of NDMAF. The network has a distinct "community" structure in the local area, including 2 national communities, 2 regional communities and 3 local-level communities.

“腾讯迁徙”大数据基于位置服务,实时、动态、完整、系统地描述了用户日常出行活动的轨迹。通过采集“腾讯迁徙”数据平台中2017年国庆-中秋长假期间国内299个城市之间的逐日人口流动数据,分“出行期、旅途期、返程期”3个时间段,利用复杂网络分析方法,从人口流动集散层级、集散网络体系的分层集聚、人口流动空间格局、网络空间特征等角度分析各时间段城市间的人口流动特征与空间格局。结果表明,腾讯迁徙大数据直观地揭示了国庆-中秋期间中国各地级城市间人口的迁移规律,3个时段人口的净流入均呈现十字形骨架支撑的菱形分布,人流集散中心主要集中在京津冀、长三角、珠三角和成渝四大城市群,与城市等级有较强的一致性。人口流动集散体系呈明显的分层集聚,城市行政级别的高低与人口流动影响力存在一定的正相关关系,大部分城市人口流动处于“相对平衡”状态。人口流动空间格局呈现出明显的核心-边缘结构,大理-鹤岗一线是人口流动强度空间分布的显著分界线,以此线为界,城市网络呈现东密西疏的分布特征和东部并联、西部串联的网络关联特征。人口流动网络总体表现出“小世界”网络特征,局部具有较明显的“社区”结构特征,聚为2个国家级、2个区域级和3个地区级社区。

With the rapid development of ICTs and big data technology, various "flow" data have become easier to be obtained such as population flows, information flows and capital flows. These data are key material to build the production network, social network, urban network and other types of network, which are becoming very important research subjects of geographical sciences and enrich the research content of this field. So far, researches on the networks in geography are still in their infancy stage. There are a number of methods that have not been utilized effectively, while it is difficult to find a proper abstracted network model to simulate the geographical phenomenon in many occasions. Therefore, to seek more appropriate methods and models for geographical network studies undoubtedly become urgent tasks in this field. In this context, an exploratory research was carried out. First, population flow network of China in the period of "Chunyun", a typical directed-weighted urban network, was built based on the population flow data acquired from “Big data on human migration during the "Chunyun" period from Baidu Map ("Baidu migration data" in short). Sequentially directed alternative centrality and directed alternative power were proposed in predecessor's foundation to evaluate the importance of each node and divide these nodes into different types according to their roles in the urban network. As analysis result, two dimensional spatial characteristic and geographical spatial characteristic was revealed which endowed the methods more practical significances. To validate the effectiveness of the methods, a contrastive analysis of the directed alternative centrality/directed alternative power and centrality/directed alternative power in previous studies was conducted. Based on directed weighted urban network and with the help of this method, this paper classified the cities into quintessential cities, hub cities, gateway cities and periphery cites, and identified the characteristics of agglomeration and diffusion at the mean time. And the result of classification was explained according to the internal structure and external context of the population flow network during the "Chunyun" period. It is found that the regional difference of directed alternative centrality and directed alternative power have certain relationship to economic and social development, national development strategy and administrative division. This explorative process demonstrated that one can effectively extract the information on the hierarchical structure, function divergence and imbalance of the directed weighted urban network via the directed alternative centrality-directed alternative power method. Directed alternative centrality-directed alternative power method is really a feasible approach to analyze the geographical network, and it also demonstrated that directed-weighted urban network can really make a difference from undirected and unweighted network in the geographical network studies.

目前中国地学领域对网络的研究仍处于起步阶段,大量网络研究方法尚未得到应用的同时,许多地理现象尚未寻找到合适的抽象网络模型,不断拓展和应用有效的网络分析方法成为当务之急。在这一背景下,基于百度迁徙平台获取的春运人口流动大数据,构建中国春运人口流动的有向加权城市网络,在已有研究基础上提出有向转变中心性与控制力测度方法,对该城市网络的节点重要性和功能进行评估和分类,分析其呈现的二维空间特征与地理空间分布特征,并与已有方法进行对比。结果表明：利用有向转变中心性和控制力方法,可有效提取有向加权网络的层级结构、功能分异和非均衡性等特征,不但可区分典型城市、枢纽城市、门户城市与边缘城市等不同的城市类型,还可以识别城市的集聚与扩散特征,从而获得比无权无向网络更为丰富和具有实践意义的信息参考,是一种有效的地理网络分析方法。

Based on Baidu migration data,this paper constructs the relationship matrix of population movement,and fully depicts the spatial agglomeration and diffusion process of inter-provincial circulation of Chinese population in the eastern,central and western regions during the spring festival of 2015．The conclusions are as follows: ( 1) February 13th-17th and February 25th-March 1st are respectively the back-home flow and the postganglionic return flow peak periods; ( 2) The inter-provincial flow is char- acterized by both near field and spatial dependency．Beijing-Tianjin-Hebei,Yangtze River Delta,and Pearl River Delta city group led the high-intensity population flows,and form three fan-shaped radiation areas; ( 3) we identify 94 population inflow cities and 242 outflow cities．The population agglomeration along the coastal areas and the population collapse in middle areas form the double vertical pattern of the floating population in China．

文章基于百度迁徙数据构建人口流动的关系矩阵,完整刻画出 2015 年春运期间全国 人口在东部与中西部城市间省际循环流动的空间集聚与扩散过程。主要研究结论: ( 1) 2015 年 2 月 13 日( 腊月廿五) 至 17 日( 腊月廿九) 和 2 月 25 日( 正月初七) 至 3 月 1 日( 正月十一) 是春节前返乡 流和节后返工流两阶段人口净流入( 流出) 的峰值时段,两阶段各市日人口净流入人次存在正负对应 的规律性特征; ( 2) 人口省际流动的近域特征、空间依赖关系并存,京津冀、长三角、珠三角城市群主导了 高强度人口流动,并形成3 个扇形辐射区域;( 3) 识别出全国 94 个人口流入城市和 242 个流出城市。在 空间上,人口集聚的沿海绵延带与人口流失的中部塌陷带共同构成流入、流出城市的双纵平行格局。

Population migration flows between different regions are related to not only the origin- and destination-specific characteristics, but also to the migration flows to and from neighborhoods. Intuitively, changes in the characteristics of a single region will impact both inflows and outflows to and from other regions. In order to explore the spatial interaction mechanism driving the increasing population migration in China, this paper builds the spatial OD model of interprovincial migration flows based on the sixth national population census data and related social-economic data. The findings are as follows: (1) Migration flows show significant autocorrelation effects among origin and destination regions, which means that the migration behavior of migrants in some region is influenced by that of migrants in other places. The positive effects indicate the outflows from an origin or the inflows to a destination tend to cluster in a similar way. Simultaneously, the negative effects suggest the flows from the neighborhood of an origin to the neighborhood of a destination tend to disperse in a dissimilar way. (2) Multilateral effects of the regional economic and social factors through the spatial network system lead to the clustering migration flows across interrelated regions. Distance decay effect plays the most influential force in shaping the patterns of migration flows among all the factors and the negative spillover effect further aggravates the friction of distance. As for destinations, the influence of wage level and migration stocks is beyond that of GDP and the positive spillover effects of these factors enhance the attraction of neighborhood regions. The spillover effects of unemployment rate and college enrollment of higher education are significantly negative while the effect of population in a destination is not significant. As for origins, population and migration stocks lead to positive spillover effects on the neighborhoods while the effects of other factors are negative. (3) Changes in the regional characteristics will potentially lead to a series of events to the whole migration system, and the flows to and from the center of oscillation and its neighborhoods vibrate greatly compared with other regions. The simulation results of 5% GDP increase in Jiangsu province indicate that the outflows to other regions decrease while the inflows from all others increase to some different extent. Comparatively, the influence on the flows to and from the regions neighboring Jiangsu is significant while that of remote regions is much less, which cannot be explained by the traditional gravity model.

区际人口迁移不仅与迁出地和目的地的要素特征以及距离有关,而且还受到周边迁移流的影响.基于网络自相关理论,利用"六普"省际人口迁移数据和相关统计资料,在重力模型的基础上考虑迁移流之间可能存在的几种空间依赖形式,构建中国省际迁移流的空间OD模型,初步揭示区域经济社会等因素及其空间溢出效应对省际人口迁移的影响,并就区域要素变化对整个省际人口迁移系统产生的"连锁反应"进行了模拟.结果表明:① 中国省际迁移流之间存在显著的网络自相关效应.目的地和迁出地的自相关效应皆为正,导致迁入和迁出流的空间效仿行为;迁出地和目的地周边则出现负的自相关效应,导致迁移流的空间竞争行为;② 区域经济社会等因素通过网络空间关系对周边地区产生的多边溢出效应导致迁移流在空间上集聚.其中,距离衰减效应位居各要素之首,其溢出效应进一步加剧距离的摩擦作用;对目的地而言,区域工资水平和迁移存量超过GDP的影响并产生正的溢出效应,促进周边地区吸引更多的外来人口;对迁出地而言,人口规模和迁移存量产生正的溢出效应,推动周边地区人口外迁;③ 区域要素变化潜在地对整个省际人口迁移系统产生一系列"连锁反应",震荡中心及其周边区域的迁移流波动较大.江苏省GDP增长5%的模拟结果表明,江苏迁往全国其他省份的人口数量都有不同程度地减少,而其他省份入迁人口均有所增加.相对而言,江苏周边省份的迁入或迁出流受到的波动较大,偏远省份波及较小,这是传统的重力模型所无法解释的.

Simulations based on spatial interaction models have been widely applied to understand the strength of relationships between geographical elements, but many issues remain unclear and deviations between actual and simulated results have often been seriously underestimated. A high-precision Baidu migration process combined with mass relationships is applied in this study and enables the generation of regression coefficients of gravity model based on programmed large-scale regression simulations. A series of accuracy assessments are then developed for 2015 empirical projection daily regression coefficients that can be applied to Chinese spring interprovincial mobile gravity model variables as well as spatiotemporal research that utilizes regression coefficients within a heterogeneity research model. This approach also enables the error within the gravity model to be assessed in terms of floating population simulations. The results of this analysis lead to a number of clear conclusions, including the fact that parameter calibration complexity for the Chinese population mobility gravity model is reflected in the degree of influence asymmetry within spatial object interaction variables, and that the spatial heterogeneity of the variable regression coefficient increases in two distinct fashions. The first of these increases has to do with the overall influence of specific variables, including the fact that differences between proxies tend to be higher than inflow-outflow characteristics. In contrast, the second set of increases is related to economic levels, industrial scales, the proportion of the tertiary industry, and public service facilities. In this latter case, two-way population flow exerts a more profound influence on results and thus the scope of possible explanations for phenomena is more extensive. The regression coefficient for the existence of positive and negative proxy variables therefore relates to differences in spatial heterogeneity, including at the city level, and also assumes that floating population gravity model regression coefficients ignore spatiotemporal changes in the heterogeneity coefficient. This leads to spatial differences in estimated results and thus convergence trends, but further enables the identification of anisotropic interactions in extension space. The second main conclusion of this research is that the national scale population flow distance attenuation coefficient was 1.970 during the spring of 2015, while at the level of prefectural administrative units and given population outflow, the range encapsulated by this coefficient fell between 0.712 (Zhumadian) and 7.699 (Urumqi). Data also reveal a population inflow coefficient of 0.792 for this year that ranged as high as 8.223 in both Sanya and Urumqi. Population flow simulation results using the gravity model and including Baidu migration measured flow data were also subject to significant error. Third, the results of this analysis reveal a total fitting error of 85.54% in weighted absolute mean; the spatial interaction effect within this is responsible for a maximum error of 86.09% in actual and simulated flows, while relative outflow force and attractiveness encompass 57.73% and 49.34% of model error, respectively. These results show that the spatial interaction effect remains most difficult to model in terms of current factors.

空间交互模型被广泛应用于地理要素关系强度的模拟,然而目前大量研究或建立在模型参数标定理想化、模式化的假设条件下,或是在暗箱中完成,由此导致模拟结果与实际的偏差却被严重低估。基于2015年中国春运期间人口省际流动的城市间O-D数据,在逐日、分市的研究精度下,实证推算人口流动重力模型变量的回归系数,探究模型代理变量影响效应的空间异质性,并评估重力模型在人口流动模拟上的误差。结果显示：① 重力模型参数标定的复杂性体现在交互对象代理变量影响程度的非对称性,和变量回归系数的空间异质性随研究精度加深显著加剧两个方面,因此模型参数标定的模式化将导致估算结果空间差异的趋势收敛;② 2015年春运期间中国人口省际流动距离衰减系数为1.970,在地级行政单元视角下,人口流出地距离衰减系数值域为0.712（驻马店）~7.699（乌鲁木齐）,人口流入地系数值域为0.792（三亚）~8.223（乌鲁木齐）;③ 应用重力模型模拟人口流动结果与实测流（百度迁徙数据）存在显著误差。就加权绝对平均误差而言,拟合总误差为85.54%,其中空间相互作用效应造成了86.09%的实测流与模拟流的最大误差,相对流出力、相对吸引力分别造成57.73%、49.34%的模型误差。因此,空间交互效应仍然是当前最难以模式化的因素。

According to previous studies, not only does the conditional gravity model based on ordinary least squares often bring about poor fitting of migration flows in reality, but also there exists overdispersion in the extended Poisson gravity model. Simultaneously, network autocorrelation usually exists in population migration data （e.g., the spatial interaction among migration flows). The problems mentioned above result in biased estimation. In order to capture network autocorrelation and deal with the issue of overdispersion, we build an eigenvector spatial filtering negative binomial gravity model (ESF NBGM) based on the data of 1% national population sample survey in 2015, to analyze the driving mechanism of interprovincial population migration flows in China. The results are as follows: (1) Positive spatial spillover effect exists in interprovincial population migration flows, and ESF can capture network autocorrelation in data, so as to reduce the estimated deviation of the model. Furthermore, eigenvectors ranking top 1.4% can properly interpret the spatial pattern of high network autocorrelation in data. (2) There exists overdispersion in China's interprovincial migration flows. Considering this problem, a negative binomial regression model is more suitable for the estimation of driving mechanism for population migration, together with statistical enhancement. (3) Network autocorrelation leads to overestimation of distance variables and underestimation of non-distance variables. The results of the improved model reveal that: chief factors the affect driving mechanism are regional population characters, social network, economic development and education level. Meanwhile, living environment and road network gradually become one of the most crucial pulling factors that influence migration flows. (4) Compared to previous studies, social network (i.e. migration stock) plays a more significant role in population migration flows, while the impact of spatial distance keeps weakening.

人口迁移数据中往往存在较强的网络自相关性,以往基于最小二乘估计的重力模型与迁移数据的拟合度较低,而改进后的泊松重力模型仍存在过度离散的缺陷,以上问题均导致既有人口迁移模型中的估计偏差。本文构建了特征向量空间滤波（ESF）负二项重力模型,基于2015年全国1%人口抽样调查数据,研究2010-2015年中国省际人口迁移的驱动因素。结果表明：① 省际人口迁移流间存在显著的空间溢出效应,ESF能有效地提取数据中的网络自相关性以降低模型的估计偏差,排序在前1.4%的特征向量即可提取较强的网络自相关信息。② 省际人口迁移流之间存在明显的过度离散现象,考虑到数据离散的负二项重力模型更适用于人口迁移驱动因素的估计。③ 网络自相关性会导致模型对距离相关变量估计的上偏与大部分非距离变量估计的下偏,修正后的模型揭示出以下驱动因素：区域人口特征、社会网络、经济发展、教育水平等因素是引发省际人口迁移的重要原因,而居住环境与公路网络等因素也逐渐成为影响人口迁移重要的“拉力”因素。④ 与既有研究相比,社会网络因素(迁移存量、流动链指数)对人口迁移的影响日益增强,而空间距离对人口迁移的影响进一步呈现弱化趋势。

"In this paper, [the authors] suggest an alternative method for fitting the gravity model. In this method, the interaction variable is treated as the outcome of a discrete probability process, whose mean is a function of the size and distance variables. This treatment seems appropriate when the dependent variable represents a count of the number of items (people, vehicles, shipments) moving from one place to another. It would seem to have special advantages where there are some pairs of places between which few items move. The argument will be illustrated with reference to data on the numbers of migrants moving in 1970-1971 between pairs of the 126 labor market areas defined for Great Britain...."excerpt

Widely available data on the number of people living outside of their country of birth do not adequately capture contemporary intensities and patterns of global migration flows. We present data on bilateral flows between 196 countries from 1990 through 2010 that provide a comprehensive view of international migration flows. Our data suggest a stable intensity of global 5-year migration flows at ~0.6% of world population since 1995. In addition, the results aid the interpretation of trends and patterns of migration flows to and from individual countries by placing them in a regional or global context. We estimate the largest movements to occur between South and West Asia, from Latin to North America, and within Africa.

Data on stocks and flows of international migration are necessary to understand migrant patterns and trends and to monitor and evaluate migration-relevant international development agendas. Many countries do not publish data on bilateral migration flows. At least six methods have been proposed recently to estimate bilateral migration flows between all origin-destination country pairs based on migrant stock data published by the World Bank and United Nations. We apply each of these methods to the latest available stock data to provide six estimates of five-year bilateral migration flows between 1990 and 2015. To assess the resulting estimates, we correlate estimates of six migration measures from each method with equivalent reported data where possible. Such systematic efforts at validation have largely been neglected thus far. We show that the correlation between the reported data and the estimates varies widely among different migration measures, over space, and over time. We find that the two methods using a closed demographic accounting approach perform consistently better than the four other estimation approaches.

China's migrant population continued to grow for decades, but has slowly declined since 2015. In such a context, the destination choice of stock migrants has a profound influence on the future pattern of population distribution and urbanisation. Previous analysis and modelling of internal migration in China have mainly focused on the primary migration or the most recent one, while it has long been recognized that migration is a repetitive event. Migrants have various motives and strategies in different stages, and the experience gained from the primary migration will affect the subsequent one. So, the primary and onward migrations differ in their spatial patterns. This paper focused on China's onward migrants who have moved once. Considering both the origins and primary destinations, we constructed a framework of how multi-proximity, which includes cognitive proximity, geographic proximity, institutional proximity, informational proximity and social proximity, which generates an influence on destination choice by affecting the migration cost. Based on data from the 2017 China Migrants Dynamic Survey, and by the use of the nested logit model, we make an empirical test. The results show that migrants have a dual identity of the origin and the primary destination, and their onward migration is based simultaneously on these two places. After controlling the effects of cities' characteristics, dual multi-proximity has a significant and robust impact on migrants' decision-making. Onward migrants prefer destinations which match them well in education background, and are geographically adjacent to, and closely connected with their origins and previous destinations. The migration network between a destination and its origins increases the likelihood that onward migrants will migrate to such a destination. Besides, migrants are willing to return to the province of their origins subjectively, but cities in such provinces are generally lack of attractiveness due to low level of socio-economic development. However, the effect of proximities varies from different groups. This is reflected in the following facts. First, women are more dependent on their hometown networks than men. Second, the new generation of migrants is similar to the older one in many ways, but relies more on the information and social connections of their hometowns than the latter. Finally, highly educated migrants are able to take advantage of localized social capital owing to their deep social integration in the primary destinations and are less dependent on the social capital in their origins. So, their destination choice is more similar to local residents than their counterparts.

中国人口流动进入总量稳定下的空间调整新阶段,存量流动人口的再流动成为塑造人口和城镇化格局的主体因素。本文从户籍地和前期流入地两个基点出发,综合考虑地理、制度、信息、社会、认知等维度,构建了双重多维邻近性影响人口再流动目的地选择的理论框架,利用全国流动人口动态监测调查数据和嵌套Logit模型开展实证研究。结果表明：双重多维邻近性对人口再流动的空间选择具有显著而稳健的影响。人口再流动时,会选择与自身知识水平更匹配、与户籍地和前期流入地都更近、信息联系都更强的城市,但社会联系仍主要依靠老家。流动人口具有回到户籍所在省（区、市）的主观意愿,但省内城市普遍机会不多、环境欠佳,缺乏吸引力。相对而言,女性对同乡网络的依赖高于男性;新生代流动人口在很多方面继承了老一代的特点,甚至比后者更依赖老家的信息和社会联系;高学历流动人口在流入地的深度社会融入使其能够利用本地化社会资本,再流动的空间选择也与流入地居民更为类似。