Village and town network clustering method

Abstract

The invention relates to a village and town network clustering method. The method comprises the following steps: obtaining, verifying and preliminarily processing mobile phone signaling data; constructing an overall village and town network based on the mobile phone signaling data; and carrying out network clustering calculation to obtain a village and town community structure. According to the method, cross-administrative-village full-purpose travel and commuting travel data of mobile phone users are identified by using mobile phone signaling to represent people flow connection between villages and towns, a village and town network is constructed by referring to a complex network analysis method, clustering calculation is performed by using a Louvain community discovery algorithm, and finally a plurality of communities and network structures in the communities are obtained. And an efficient method is provided for positioning and optimizing the space organization and development of the village and town settlement and village and town system in China and realizing the optimal configuration of land space resources.

Description

Technical Field

[0001] The present invention relates to the field of network clustering, and in particular to a village and town network clustering method. Background Art

[0002] With the continuous advancement of urbanization in China, the relationship between villages and cities has become closer, and they have gradually participated in the process of urban industrial division of labor and production specialization; the functional differentiation between different village and town settlements and the fuzziness of the village and town system hierarchy have become increasingly obvious, and the traditional spatial organization logic based on the central place theory is no longer fully applicable to village and town settlements. At present, there is a relative lack of analysis technology for village and town settlements, and the traditional data sample size is small, difficult to obtain, and timeliness is poor. As a result, my country's village and town system planning often lacks scientific data support, resulting in unreasonable village and town division, and thus problems such as spatial resource mismatch, public infrastructure, and imbalance in service supply and demand. Summary of the invention

[0003] In view of this, the present invention provides a village and town network clustering method, which effectively solves the current problem of being unable to provide scientific and accurate data support for the new village and town system planning, and provides new technical methods for my country's village and town system planning, village and town spatial layout planning and construction fields.

[0004] In order to achieve the above object, the present invention provides a village and town network clustering method, comprising the following steps:

[0005] S1. Acquisition, verification and preliminary processing of mobile phone signaling data;

[0006] S2, build the overall network of villages and towns based on mobile phone signaling data;

[0007] S3. Network clustering calculation to obtain the village and town community structure.

[0008] Preferably, the acquisition, verification and preliminary processing of mobile phone signaling data include the following steps:

[0009] S101, dividing the acquired mobile phone signaling data into two types of data: all-purpose travel and commuting travel according to the location and duration of the mobile phone user's stay;

[0010] S102: The base map data adopts the shp format file with administrative villages as the basic spatial unit. Generally, the township areas of cities and counties are selected and the county towns and urban areas are excluded;

[0011] S103, the population identified by the signal base station is divided into the population of the administrative village where the spatial location of the signal base station is located, and based on this, the obtained mobile phone signaling data is preliminarily processed through the PostgreSQL database management system and ArcGIS platform to obtain the data of population travel connections between administrative villages;

[0012] S104. Perform regression analysis on the statistical permanent population of each district, county, town and street in the research area and the resident population obtained through mobile phone signaling. The goodness of fit R 2 The higher it is, the more accurate and reliable the mobile phone signaling data is; the number of residents obtained through mobile phone signaling within the research scope is divided by the statistical permanent population of each district, county, town and street to obtain the mobile phone signaling population recognition rate of each district, county, town and street. If the recognition rate is too low, it will affect the accuracy and authenticity of the final calculation result.

[0013] Preferably, building a village and town overall network based on mobile phone signaling data includes the following steps:

[0014] S201. The center point of the administrative village is used as the node of the network, and the population travel connection between administrative villages is used as the edge. Specifically, the frequency of all-purpose travel is used as the edge weight, and then the number of commuting trips is used as the edge weight. The number of people flowing in both directions is accumulated respectively, thereby establishing two undirected networks. The formula is as follows:

[0015] W ij =C ij +C ji

[0016] Among them, i and j are nodes, W ij is the weight of the edge between node i and node j, C ij is the number of people traveling from node i to node j, C ji is the number of people traveling from node j to node i;

[0017] S202. Assign edge weights to the lines between the village and town points on the ArcGIS platform to achieve visualization of the village and town network.

[0018] Preferably, the network clustering calculation adopts Louvain (BGLL) algorithm, and the Louvain (BGLL) algorithm includes the following steps:

[0019] S301, substituting the processed village and town network data into the calculation process of igraph package and Louvain algorithm network clustering;

[0020] S302, first consider each vertex as an independent community, traverse each vertex in a certain order, and the change in modularity when vertex m moves to its neighbor vertex n is △Q; if △Q>0, move vertex m to the administrative village of the vertex with the largest change; if △Q<0, vertex m remains unchanged; repeat this process until the movement of each vertex cannot increase the modularity, and obtain several new community structures after the first iteration;

[0021] S303, treating each community obtained in step S302 as a new independent vertex, repeating the process of step S302 to start a new round of iteration; after each iteration, the communities generated in the previous iteration are merged, and the number of communities is reduced, until all nodes in the entire graph no longer change and the modularity no longer increases, then the iteration ends;

[0022] S304: Visualize the community structure obtained by clustering and the network structure within the community on the ArcGIS platform.

[0023] The beneficial effects of the present invention are:

[0024] (1) The present invention uses mobile phone signaling to identify the all-purpose travel and commuting travel data of mobile phone users across administrative villages to represent the flow of people between villages and towns. Compared with other traditional data, the method of obtaining data through mobile phone signaling is relatively simple, has a wide coverage, a large number of samples, and is real-time and sustainable. The huge database provides accurate data for clustering calculations.

[0025] (2) The Louvain (BGLL) algorithm used in the present invention is a heuristic algorithm based on modularity optimization. The algorithm has two layers of iteration. The outer iteration is a bottom-up condensation method, and the inner iteration is a condensation method plus an exchange strategy. This avoids the shortcoming of a simple condensation method that two nodes cannot be separated once they are merged.

[0026] (3) The analysis results of the network clustering of the present invention are relatively reliable and can clearly and intuitively reflect the spatial structural characteristics of the village-town system, which is conducive to supporting the development planning and spatial optimization of the village-town system under the background of national land space planning, and provides favorable assistance for the development of public infrastructure planning practices such as rural community living circles. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flow chart of the present invention;

[0028] Figure 2 This is the distribution map of China Mobile base stations in Hangzhou;

[0029] Figure 3 This is the recognition rate and regression verification diagram of the mobile phone signaling resident population in each district and county of Hangzhou;

[0030] Figure 4 The network of villages and towns in Hangzhou based on commuting connections;

[0031] Figure 5 The network of villages and towns based on commuting connections in Lin'an District, Hangzhou;

[0032] Figure 6 (a), (b), and (c) are the community structure diagrams of the three iterations of the village-town network clustering based on commuting travel connections in Hangzhou;

[0033] Figure 7 This is the community and intra-community network structure diagram obtained from the first iteration of village-town network clustering based on commuting travel connections in Lin'an District, Hangzhou. DETAILED DESCRIPTION

[0034] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation mode, structure, characteristics and effects of the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments.

[0035] This embodiment discloses a village and town network clustering method, comprising the following steps:

[0036] S1. Acquisition, verification and preliminary processing of mobile phone signaling data;

[0037] S2, build the overall village network based on mobile phone signaling data;

[0038] S3. Network clustering calculation to obtain the village and town community structure.

[0039] The all-purpose travel and commuting travel data of mobile phone users across administrative villages identified by mobile phone signaling represent the flow of people between villages. The acquisition, verification and preliminary processing of mobile phone signaling data include the following steps:

[0040] S101, dividing the acquired mobile phone signaling data into two types of data: all-purpose commuting and commuting travel according to the location and duration of the mobile phone user's stay;

[0041] S102: The base map data adopts the shp format file with administrative villages as the basic spatial unit. Generally, the township areas of cities and counties are selected and the county towns and urban areas are excluded;

[0042] S103, the population identified by the signal base station is divided into the population of the administrative village where the spatial location of the signal base station is located, and based on this, the obtained mobile phone signaling data is preliminarily processed through the PostgreSQL database management system and ArcGIS platform to obtain the data of population travel connections between administrative villages;

[0043] S104. Perform regression analysis on the statistical permanent population of each district, county, town and street in the research area and the resident population obtained through mobile phone signaling. The goodness of fit R 2 The higher the value, the more accurate and reliable the mobile phone signaling data is. Figure 2 , Figure 3As shown in the figure; the statistical permanent population of each district, county, town and street within the research scope is divided by the number of residents obtained through mobile phone signaling to obtain the mobile phone signaling population recognition rate of each district, county, town and street. If the recognition rate is too low, it will affect the accuracy and authenticity of the final calculation result; there are some administrative villages without base stations, which is almost non-existent in China and can be ignored or the administrative village unit can be appropriately merged before calculation.

[0044] Building a village network based on mobile phone signaling data includes the following steps:

[0045] S201. The center point of the administrative village is used as the node of the network, and the population travel connection between administrative villages is used as the edge. Specifically, the frequency of all-purpose travel is used as the edge weight, and then the number of commuting trips is used as the edge weight. The number of people flowing in both directions is accumulated respectively, thereby establishing two undirected networks. The formula is as follows:

[0046] W ij =C ij +C ji

[0047] Among them, i and j are nodes, W ij is the weight of the edge between node i and node j, C ij is the number of people traveling from node i to node j, C ji is the number of people traveling from node j to node i;

[0048] S202, assign edge weights to the links between village and town points on the ArcGIS platform to achieve visualization of the village and town network, such as Figure 4 , Figure 5 shown.

[0049] The network clustering calculation adopts the Louvain (BGLL) algorithm, which includes the following steps:

[0050] S301, substituting the processed village and town network data into the calculation process of igraph package and Louvain algorithm network clustering;

[0051] The network data must be processed as having no multiple edges or loop edges. Multiple edges refer to edges with two or more vertices being the same. For example, if there is more than one A loop edge is an edge with two or more vertices opposite to each other. and If there are multiple edges or loop edges between two edges, their weight values ​​should be merged in advance.

[0052] For the same set of data, if the order of each row of data is changed, it may cause differences in the iteration process of the middle class. This is because the order of the clustering starting points has changed, but this has little effect on the final iteration result.

[0053] S302, first consider each vertex as an independent community, traverse each vertex in a certain order, and the change in modularity when vertex m moves to its neighbor vertex n is △Q; if △Q>0, move vertex m to the community of the vertex with the largest change; if △Q<0, vertex m remains unchanged; repeat this process until the movement of each vertex cannot increase the modularity, and obtain several new community structures after the first iteration;

[0054] The increase in modularity is to make the subcommunity have more internal edges and fewer external edges, that is, more clustered.

[0055] S303, treating each community obtained in step S302 as a new independent vertex, repeating the process of step S302 to start a new round of iteration; after each iteration, the communities generated in the previous iteration are merged, and the number of communities is reduced, until all nodes in the entire graph no longer change and the modularity no longer increases, then the iteration ends;

[0056] Each iteration will result in several different and new community structures. The further the iteration goes, the more communities will be merged. Therefore, the number of communities obtained is smaller and the number of nodes contained in the communities is more. As shown in Figure 6, the modularity of the village network clustering based on commuting travel connections in Hangzhou has exceeded 0.7 after the first iteration, and the modularity has exceeded 0.85 after the second iteration, which is a significant improvement compared with the first iteration, indicating that the internal agglomeration degree of the community is high. The modularity of the third iteration is relatively stable compared with the second iteration, which shows that the community structure formed is relatively stable and the results are more reliable.

[0057] S304: Visualize the community structure obtained by clustering and the network structure within the community on the ArcGIS platform.

[0058] Village and town network clustering technology is mainly used in the fields of village and town system planning, village and town spatial layout planning and construction in my country; the application scope of village and town network clustering technology can be determined according to the population size of the administrative village / community, the level of economic and social development, the natural geographical conditions, the distribution of base stations and the quality of mobile phone data acquisition. It can cover the entire township scope of the city or county, and the research spatial unit is the administrative village / community level; in the application process, the overall and local structural characteristics of the village and town network from the perspective of human flow are considered to optimize the spatial organization and development positioning of village and town settlements and village and town systems, and achieve optimal allocation of national space resources.

[0059] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent embodiments without departing from the scope of the technical solution of the present invention. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A village and town network clustering method, characterized in that: The following steps are involved: S1. Acquisition, verification and preliminary processing of mobile phone signaling data; S2, build the overall network of villages and towns based on mobile phone signaling data; S3. Network clustering calculation to obtain the village and town community structure.

2. A village and town network clustering method according to claim 1, characterized in that: The acquisition, verification and preliminary processing of the mobile phone signaling data include the following steps: S101, dividing the acquired mobile phone signaling data into two types of data: all-purpose travel and commuting travel according to the location and duration of the mobile phone user's stay; S102: The base map data adopts the shp format file with administrative villages as the basic spatial unit. Generally, the township areas of cities and counties are selected and the county towns and urban areas are excluded; S103, the population identified by the signal base station is divided into the population of the administrative village where the spatial location of the signal base station is located, and based on this, the obtained mobile phone signaling data is preliminarily processed through the PostgreSQL database management system and ArcGIS platform to obtain the data of population travel connections between administrative villages; S104. Perform regression analysis on the statistical permanent population of each district, county, town and street in the research area and the resident population obtained through mobile phone signaling. The goodness of fit R 2 The higher it is, the more accurate and reliable the mobile phone signaling data is; the number of residents obtained through mobile phone signaling within the research scope is divided by the statistical permanent population of each district, county, town and street to obtain the mobile phone signaling population recognition rate of each district, county, town and street. If the recognition rate is too low, it will affect the accuracy and authenticity of the final calculation result.

3. A village and town network clustering method according to claim 1, characterized in that: The construction of the overall village and town network based on mobile phone signaling data includes the following steps: S201. The center point of the administrative village is used as the node of the network, and the population travel connection between administrative villages is used as the edge. Specifically, the frequency of all-purpose travel is used as the edge weight, and then the number of commuting trips is used as the edge weight. The number of people flowing in both directions is accumulated respectively, thereby establishing two undirected networks. The formula is as follows: W ij =C ij +C ji Among them, i and j are nodes, W ij is the weight of the edge between node i and node j, C ij is the number of people traveling from node i to node j, C ji is the number of people traveling from node j to node i; S202. Assign edge weights to the lines between the village and town points on the ArcGIS platform to achieve visualization of the village and town network.

4. A village and town network clustering method according to claim 1, characterized in that: The network clustering calculation adopts Louvain (BGLL) algorithm, and the Louvain (BGLL) algorithm includes the following steps: S301, substituting the processed village and town network data into the calculation process of igraph package and Louvain algorithm network clustering; S302, first consider each vertex as an independent community, traverse each vertex in a certain order, and the change in modularity when vertex m moves to its neighbor vertex n is △Q; if △Q>0, move vertex m to the community of the vertex with the largest change; if △Q<0, vertex m remains unchanged; repeat this process until the movement of each vertex cannot increase the modularity, and obtain several new community structures after the first iteration; S303, treating each community obtained in step S302 as a new independent vertex, repeating the process of step S302 to start a new round of iteration; after each iteration, the communities generated in the previous iteration are merged, and the number of communities is reduced, until all nodes in the entire graph no longer change and the modularity no longer increases, then the iteration ends; S304: Visualize the community structure obtained by clustering and the network structure within the community on the ArcGIS platform.

Images