A method, device, medium and product for verifying a permanent population identification result

Abstract

The application discloses a permanent population identification result verification method and device, medium and product, relates to the technical field of geographic information science and urban population analysis, and comprises the following steps: acquiring a to-be-verified permanent population identification result dataset identified based on mobile phone signaling data, individual-level conservative reference data and aggregated-level reference data; calculating an individual-level verification index according to the to-be-verified permanent population identification result dataset and the individual-level conservative reference data; calculating an aggregated-level verification index of each spatial scale under different spatial scales; obtaining a multi-scale spatial pattern verification index under a street scale and a grid scale; and collecting the individual-level verification index, the aggregated-level verification index and the multi-scale spatial pattern verification index. The application realizes full-dimensional verification from an individual to a city and from a statistical value to a spatial pattern, and improves the accuracy and efficiency of verification.

Description

Technical Field

[0001] This application relates to the fields of geographic information science and urban population analysis technology, and in particular to a method, device, medium and product for verifying the results of resident population identification. Background Technology

[0002] With the widespread application of mobile signaling data in urban population analysis, the reliability assessment of resident population identification algorithms has become a key technical challenge. Related verification methods suffer from the following significant technical deficiencies: 1. Lack of individual-level verification capability: Most studies can only conduct verification at the macro-aggregate level such as districts, counties, and streets, which cannot achieve large-scale, refined quantitative evaluation of the accuracy of individual resident identification, resulting in the inability to guarantee the reliability of individual identification accuracy.

[0003] 2. Incomplete verification scale: A multi-scale, systematic verification framework has not yet been formed, ranging from individuals to cities and from micro-grids to macro-administrative units, making it difficult to cover the verification needs of different application scenarios.

[0004] 3. Limited spatial verification dimensions: Aggregate-level verification relies heavily on simple numerical correlation analysis, ignoring the inherent spatial autocorrelation characteristics of population distribution, such as spatial clustering and diffusion, and cannot comprehensively assess the authenticity of the spatial patterns in the identification results. Summary of the Invention

[0005] The purpose of this application is to provide a method, device, medium and product for verifying the results of permanent resident identification, which can realize individual-level verification, comprehensive verification scale and multi-dimensional spatial verification.

[0006] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a method for verifying the identification results of permanent residents, including: The system acquires a dataset of resident population identification results to be verified based on mobile signaling data, individual-level conservative benchmark data, and aggregate-level benchmark data. The resident population identification result dataset includes a user set and the attribute information of the user set. The aggregate-level benchmark data includes official statistical resident population data, total number of users in the operator market, and high-resolution raster population dataset. The high resolution range is less than or equal to 500m.

[0007] The unweighted and weighted permanent resident population numbers are calculated based on the dataset of permanent resident identification results to be verified at multiple spatial scales; the spatial scales include: city scale, district / county scale, street scale, and grid scale.

[0008] Baseline population aggregation values ​​at multiple spatial scales are calculated based on high-resolution raster population datasets.

[0009] The dataset of permanent resident identification results to be verified is matched with individual-level conservative benchmark data using user unique identifiers to calculate individual-level verification indicators.

[0010] At different spatial scales, aggregate-level verification indicators are calculated for each spatial scale based on unweighted permanent resident population, weighted permanent resident population, and aggregate-level baseline data.

[0011] Global spatial autocorrelation and cluster analysis were performed on the unweighted permanent resident population, weighted permanent resident population, and baseline population aggregation value at both the street and grid scales to obtain multi-scale spatial pattern verification indicators.

[0012] Individual-level verification metrics, aggregate-level verification metrics, and multi-scale spatial model verification metrics are summarized.

[0013] Secondly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the verification method for the identification results of permanent residents described above.

[0014] Thirdly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the verification method for the identification results of permanent residents as described above.

[0015] Fourthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the verification method for the identification results of permanent residents as described above.

[0016] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method, device, medium, and product for verifying resident population identification results. It acquires a dataset of resident population identification results to be verified, individual-level conservative benchmark data, and aggregate-level benchmark data identified based on mobile phone signaling data. Based on the dataset of resident population identification results to be verified, it calculates unweighted and weighted resident population numbers at multiple spatial scales, improving the reliability of aggregate-level estimation. It also calculates benchmark population aggregation values ​​at multiple spatial scales based on a high-resolution raster population dataset, providing a benchmark reference for spatial pattern verification. Furthermore, it matches the dataset of resident population identification results to be verified with the individual-level conservative benchmark data using unique user identifiers to calculate individual-level verification indicators, achieving individual-level verification based on conservative benchmarks and filling a technological gap. At the same spatial scale, aggregated verification indicators are calculated for each spatial scale based on unweighted resident population, weighted resident population, and aggregated baseline data, constructing a multi-scale systematic verification framework covering all spatial granularities. At the street and grid scales, global spatial autocorrelation and cluster analysis are performed on the unweighted resident population, weighted resident population, and aggregated baseline population values ​​to obtain multi-scale spatial pattern verification indicators. This introduces spatial pattern verification to compensate for the lack of dimensions in numerical verification. Individual-level verification indicators, aggregated verification indicators, and multi-scale spatial pattern verification indicators are summarized for easy comparison and quantification, achieving full-dimensional verification from individuals to cities and from statistical values ​​to spatial patterns. This improves the accuracy and efficiency of verification and enables reliability assessment. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a method for verifying the identification results of permanent residents, as provided in an embodiment of this application.

[0019] Figure 2 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] In addition, the relevant verification methods have another significant technical flaw: the individual-level benchmark data is unreliable. Due to privacy protection policies and data access restrictions, it is difficult to obtain large-scale, real individual residence tags. The "home location" data provided by telecommunications operators is conservative in its updates (only updated when the cumulative nighttime stay at the new location exceeds that of the original location) and has a time lag, so it cannot be used directly as the ground truth value, resulting in a lack of reliable benchmarks for individual-level verification.

[0023] This application aims to overcome the technical deficiencies in related technologies, such as lack of individual-level verification, single spatial verification dimension, failure to systematically evaluate the authenticity of spatial patterns at multiple spatial scales, unreliable individual-level benchmark data, and imprecise statistical amplification logic. It provides a verification method for resident population identification results, which performs multi-dimensional and multi-scale systematic verification of resident population identification results based on mobile phone signaling data. Through statistical amplification weighting and multi-stage spatial attribution statistics, it achieves full-dimensional and multi-scale verification from individuals to cities and from statistical values ​​to spatial patterns, providing accurate quantitative evaluation basis for the reliability of resident population identification algorithms.

[0024] It should be stated that all data used in this application, including the dataset of resident population identification results to be verified and the multi-source benchmark dataset (i.e., individual-level conservative benchmark data and aggregate-level benchmark data), have been legally collected by telecommunications operators and data providers in accordance with the Personal Information Protection Law of the People's Republic of China and other relevant laws and regulations, with the informed consent of the users. All data has been anonymized or de-identified and cannot be directly identified or restored to the identity of a specific natural person. The use of the above data is limited to public management and scientific research purposes to improve the accuracy of urban population statistics and spatial planning, and complies with relevant legal provisions.

[0025] like Figure 1 As shown, a method for verifying the identification results of permanent residents is provided, including the following steps 201 to 207.

[0026] Step 201: Obtain the resident population identification result dataset, individual-level conservative benchmark data, and aggregate-level benchmark data identified based on mobile signaling data. The resident population identification result dataset includes user sets and attribute information of the user sets. The aggregate-level benchmark data includes official statistical resident population data, total number of users in the operator market, and high-resolution raster population dataset. The high-resolution range is less than or equal to 500m.

[0027] Step 202: Calculate the unweighted number of permanent residents and the weighted number of permanent residents at multiple spatial scales based on the permanent resident identification result dataset to be verified; the spatial scales include: city scale, district / county scale, street scale, and grid scale.

[0028] Step 203: Calculate the baseline population aggregation value at multiple spatial scales based on the high-resolution raster population dataset.

[0029] Step 204: Match the dataset of permanent resident identification results to be verified with the individual-level conservative benchmark data using the user's unique identifier, and calculate the individual-level verification index.

[0030] Step 205: At different spatial scales, calculate the aggregation-level verification index for each spatial scale based on the unweighted permanent resident population, the weighted permanent resident population, and the aggregation-level baseline data.

[0031] In another exemplary embodiment of this application, the individual-level verification metric includes recognition accuracy, and the formula for calculating the recognition accuracy is as follows: .

[0032] in, To improve recognition accuracy, This refers to the number of users in the overlapping user set obtained from the dataset of resident population identification results to be verified and the individual-level conservative benchmark data. This represents the total number of permanent residents in the dataset of permanent resident identification results to be verified.

[0033] Step 206: At the street scale and grid scale, global spatial autocorrelation analysis and cluster analysis are performed on the unweighted permanent resident population, the weighted permanent resident population, and the baseline population aggregation value, respectively, to obtain multi-scale spatial pattern verification indicators.

[0034] In another exemplary embodiment of this application, the aggregated verification indicators for each spatial scale include: the absolute relative error of the unweighted permanent resident population at the city scale, the absolute relative error of the weighted permanent resident population at the city scale, the absolute relative error of the weighted permanent resident population at the district / county scale, the Spearman correlation coefficient of the unweighted permanent resident population at the district / county scale, the Spearman correlation coefficient of the weighted permanent resident population at the district / county scale, the Pearson correlation coefficient of the unweighted permanent resident population at the district / county scale, the Pearson correlation coefficient of the weighted permanent resident population at the district / county scale, the Spearman correlation coefficient of the weighted permanent resident population at the street scale, the Pearson correlation coefficient of the weighted permanent resident population at the street scale, the Spearman correlation coefficient of the weighted permanent resident population at the grid scale, and the Pearson correlation coefficient of the weighted permanent resident population at the grid scale.

[0035] In another exemplary embodiment of this application, the multi-scale spatial pattern verification index includes: the Global Moran's I index of the unweighted resident population of the street-scale spatial pattern, the Global Moran's I index of the weighted resident population of the street-scale spatial pattern, the Global Moran's I index of the baseline population aggregation value of the street-scale spatial pattern, the General G index of the unweighted resident population of the street-scale spatial pattern, the General G index of the weighted resident population of the street-scale spatial pattern, the General G index of the baseline population aggregation value of the street-scale spatial pattern, the Global Moran's I index of the unweighted resident population of the grid-scale spatial pattern, the Global Moran's I index of the weighted resident population of the grid-scale spatial pattern, the Global Moran's I index of the baseline population aggregation value of the grid-scale spatial pattern, the General G index of the unweighted resident population of the grid-scale spatial pattern, the General G index of the weighted resident population of the grid-scale spatial pattern, and the General G index of the baseline population aggregation value of the grid-scale spatial pattern.

[0036] Step 207: Summarize the individual-level verification indicators, the aggregate-level verification indicators, and the multi-scale spatial model verification indicators.

[0037] In another exemplary embodiment of this application, the summarized individual-level verification metrics, aggregate-level verification metrics, and multi-scale spatial pattern verification metrics can be presented in the form of standardized tables or charts.

[0038] By implementing steps 201 to 207 above, this application obtains a dataset of resident population identification results to be verified, individual-level conservative benchmark data, and aggregate-level benchmark data identified based on mobile signaling data. Based on the dataset of resident population identification results to be verified, it calculates the unweighted and weighted resident population numbers at multiple spatial scales, improving the reliability of the aggregate-level estimation. Based on the high-resolution raster population dataset, it calculates the benchmark population aggregation value at multiple spatial scales, providing a benchmark reference for spatial pattern verification. The dataset of resident population identification results to be verified is matched with the individual-level conservative benchmark data using unique user identifiers to calculate individual-level verification indicators, achieving individual-level verification based on conservative benchmarks and filling a technological gap. At different spatial scales, Based on unweighted resident population, weighted resident population, and aggregated baseline data, aggregated verification indicators at each spatial scale are calculated to construct a multi-scale systematic verification framework covering all spatial granularities. At the street and grid scales, global spatial autocorrelation and clustering analyses are performed on the unweighted resident population, weighted resident population, and aggregated baseline population values ​​to obtain multi-scale spatial pattern verification indicators. This spatial pattern verification compensates for the lack of dimensions in numerical verification. Individual-level verification indicators, aggregated verification indicators, and multi-scale spatial pattern verification indicators are summarized for easy comparison and quantification. This achieves full-dimensional verification from individuals to cities and from statistical values ​​to spatial patterns, improving the accuracy and efficiency of verification and enabling reliability assessment.

[0039] In another exemplary embodiment of this application, the calculation of the unweighted number of permanent residents and the weighted number of permanent residents at multiple spatial scales based on the dataset of permanent resident identification results to be verified includes the following steps 301 to 303.

[0040] Step 301: For each geographic unit under multiple spatial scales, a two-stage spatial filtering strategy of coarse filtering and fine filtering is used to determine the residential location of the user in the resident population identification result dataset to be verified to belong to the attribution user set of each geographic unit.

[0041] Step 302: Obtain the unweighted resident population of each geographic unit from the length of the attribution user set.

[0042] Step 303: The weighted resident population of each geographic unit is obtained by summing the statistical expansion weights of all users in the attribution user set.

[0043] In another exemplary embodiment of this application, step 301 is replaced by steps 401 to 404.

[0044] Step 401: For each geographic unit at multiple spatial scales, calculate the Geohash encoding set covered by the geographic unit.

[0045] Step 402: Calculate the Geohash code corresponding to the latitude and longitude coordinates of the user's residence location based on the latitude and longitude coordinates of the user's residence location in the dataset of permanent resident identification results to be verified.

[0046] Step 403: Filter users whose Geohash codes corresponding to the latitude and longitude coordinates of their residential location belong to the set of Geohash codes covered by the geographic unit to obtain a candidate user set.

[0047] Step 404: Based on the candidate user set, perform GIS spatial intersection analysis on the latitude and longitude coordinates of the candidate users' residential locations and the geographic units to determine the user set whose residential locations are located within the geographic units, which will be used as the attribution user set.

[0048] In another exemplary embodiment of this application, step 203 is replaced by steps 501 to 502.

[0049] Step 501: For each geographic unit at multiple spatial scales, determine the set of raster cells covered by each geographic unit in the high-resolution raster population dataset.

[0050] Step 502: Extract the population value of each raster cell in the raster cell set and sum them to obtain the baseline population aggregation value of each geographic unit.

[0051] This application aims to address the shortcomings of related technologies, including the lack of individual-level verification, limited spatial verification dimensions, incomplete multi-scale verification, and unreliable individual-level benchmark data. Specifically, it includes: acquiring a dataset of resident populations to be verified (including user identifiers, residential locations, and statistical expansion weights) and multi-source benchmark data; obtaining high-confidence overlapping samples through matching with individual-level conservative benchmark data, and calculating the lower bound of individual-level identification accuracy; achieving multi-scale aggregate statistical verification using spatial attribution and weighted summation; conducting multi-scale spatial pattern verification using the Global Moran's I index and the General G index; and finally outputting a standardized quantitative indicator set. This application achieves full-dimensional verification from individuals to cities and from statistical values ​​to spatial patterns, improving the accuracy and efficiency of verification, and can be used for reliability evaluation of resident population identification algorithms.

[0052] The verification method for the permanent resident identification results in this application specifically includes the following steps.

[0053] Step 1: Obtain the multi-source benchmark dataset and the dataset of resident population identification results to be verified.

[0054] Step 1.1: Obtain the dataset of permanent resident identification results to be verified.

[0055] Obtain the dataset of resident population identification results to be verified. This dataset contains a set of users within the study area and their attribute information. The attribute information includes: a unique user identifier (used for cross-dataset matching); and identified residential location information, including the latitude and longitude coordinates of the residential location. ; Statistical expansion weight ( ): Used to characterize the user's representativeness to the total population (or a coefficient used to characterize the user's representativeness to the total population), used for weighted calculations of aggregated population estimates, and can be provided by the telecommunications operator or generated through demographic model calibration.

[0056] Step 1.2: Obtaining multi-source benchmark datasets.

[0057] The multi-source benchmark dataset includes individual-level conservative benchmark data and aggregate-level benchmark data, as detailed below.

[0058] Individual-level conservative baseline data: "Home location" tag data provided by telecommunications operators based on mobile signaling data, containing the user's unique identifier and corresponding home location information. This home location is determined based on the location with the longest cumulative nighttime stay (e.g., from 22:00 on the current day to 6:00 the next day) since the SIM card was activated, and the operator only updates the "home location" record when the user's cumulative nighttime stay in the new location exceeds that of the original location. It should be noted that this data is not a complete and current label of the individual's real residence, but rather a conservative estimate reflecting the user's long-term historical stable residence location, and the stability of the residence location has been verified through long-term nighttime stay behavior.

[0059] Aggregated benchmark data: covering multiple spatial scales including cities, districts, streets, and grids (100m-500m), including three types of data: ① official statistical resident population data (such as annual resident population census data, sampling survey data); ② the total number of market users of telecommunications operators in the region; ③ spatial high-resolution raster population datasets (such as WorldPop, LandScan, etc.).

[0060] Step 2: Calculate intermediate data.

[0061] For each geographic unit at multiple spatial scales (city scale, district / county scale, street scale, grid scale), the statistical values ​​of the resident population identification result dataset to be verified and the baseline aggregated values ​​of the high-resolution raster population dataset are calculated.

[0062] Step 2.1: For multiple spatial scales (city scale, district / county scale, street scale, grid scale), a two-stage statistical method of "spatial attribution-weighted summation" is adopted to calculate the statistical values ​​of the resident population identification result dataset to be verified, and obtain the unweighted resident population and weighted resident population of each geographical unit. The specific content is as follows.

[0063] Phase 1: Spatial Attribution (i.e., GIS-based Spatial Overlay Analysis): For each geographic unit at multiple spatial scales (including city scale, district / county scale, street scale, and grid scale), a two-stage spatial screening strategy of "coarse filtering-fine filtering" is used to determine the target user set (or attribution user set) whose residential location in the resident identification result dataset to be verified belongs to that geographic unit.

[0064] (1) Coarse filtering stage: For each geographic unit, calculate the Geohash code set covered by the geographic unit; based on the latitude and longitude coordinates of the residential location of the user in the dataset of the permanent resident identification results to be verified, calculate the corresponding 6-level Geohash code, and filter out the users whose codes belong to the Geohash code set covered by the geographic unit to obtain the candidate user set, which greatly reduces the amount of data processing.

[0065] Optionally, the coarse filtering stage filters potential users by the intersection relationship between the 6-level Geohash grid code and geographic units, with a data reduction ratio of no less than 80%.

[0066] (2) Fine filtering stage: Based on the candidate user set, the latitude and longitude coordinates of the residential location of each candidate user are filtered out one by one. By performing GIS spatial intersection analysis with geographic units, the user set whose residential location is entirely within the geographic unit is ultimately determined, which is the attribution user set of that geographic unit.

[0067] Phase 2: Based on the attribution user set of each unit, count the unweighted resident population and the weighted resident population of each geographical unit.

[0068] (1) Unweighted resident population: The length of the attribution user set for this geographic unit l .

[0069] (2) Weighted resident population: The statistical expansion weight of all users in the attribution user set of this geographic unit ( The sum of ) is calculated using the following formula: .

[0070] in, For the first Weighted resident population of each geographical unit For the first Statistical expansion weighting for each user; l The length of the attribution user set for this geographic unit.

[0071] Step 2.2: For multiple spatial scales (city scale, district / county scale, street scale, grid scale), perform aggregation operations to calculate the aggregated values ​​of the high-resolution raster population dataset and obtain the baseline aggregated population value for each geographic unit. The aggregation operations are as follows.

[0072] For each geographic unit, determine the set of raster cells it covers.

[0073] Extract the population value for each raster cell in the set.

[0074] Sum the above population values ​​to obtain the baseline aggregated population value for this geographic unit. The calculation formula is: .

[0075] in, For the first Baseline population aggregation value for each geographic unit, For the first Within the first geographical unit Population value of a raster cell This represents the number of all raster cells located within this unit.

[0076] Step 3: Individual-level verification.

[0077] Step 3.1: User matching.

[0078] Users in the dataset of permanent resident identification results to be verified are precisely matched with users in the individual-level conservative baseline data using unique user identifiers, and overlapping user sets that exist in both datasets are filtered out. The overlapping user set is a high-confidence positive sample, and the long-term stability of its residential location has been verified by the operator's conservative update rules (i.e., reflecting the user's long-term historical stable residential location), which can serve as a reliable benchmark for individual-level verification.

[0079] Step 3.2: Accuracy calculation.

[0080] Individual-level recognition accuracy is calculated using the following formula: .

[0081] in, This represents the number of users in the overlapping user set. This represents the total number of permanent residents identified in the dataset of permanent resident identification results to be verified (i.e., the total number of permanent residents in the dataset of permanent resident identification results to be verified). This accuracy rate is a conservative but reliable lower bound estimate for individual-level verification. Since the conservative individual-level benchmark data may miss some actual permanent residents due to update lag, the true accuracy rate of the permanent resident identification results is not lower than this calculated value.

[0082] Step 4: Aggregate-level statistical validation.

[0083] Step 4.1: City-scale verification. City-scale verification compares the unweighted count with the total number of users in the operator's market, and the weighted resident population with the official resident population statistics; district and county-level verification conducts correlation analysis between the unweighted count and the weighted resident population and the official statistics.

[0084] (1) Verification of unweighted permanent resident population, the calculation formula is: .

[0085] in, This represents the absolute relative error of the unweighted resident population figure. This represents the unweighted resident population at the city level. This represents the total number of city-level operator market users in the aggregated benchmark data.

[0086] (2) Verification of weighted permanent resident population, the calculation formula is: .

[0087] in, This represents the absolute relative error of the weighted resident population estimate (or weighted resident population count). This represents the weighted resident population at the city scale. This refers to the city-level official statistical resident population data in the aggregated baseline data.

[0088] Step 4.2: Validation at the district / county level.

[0089] (1) Error analysis: The absolute and relative errors of the weighted estimates for each district and county are calculated using the following formula: .

[0090] in, For the first The absolute and relative errors of each district and county For the first The weighted resident population of each district and county. For the first Official resident population statistics for each district / county. Output the ARE value sequence for each district / county, and further calculate its mean, minimum, and maximum values.

[0091] (2) The Spearman correlation coefficient was used to assess the ranking consistency between the unweighted and weighted permanent resident populations at the county / district level and the official statistics. The calculation formula is as follows: .

[0092] .

[0093] in, The Spearman correlation coefficient is... For the first The number of permanent residents in each district / county, which can be either unweighted or weighted. For the first Official statistics on the permanent resident population of each district and county. and They are respectively , Rank among all districts and counties d The number of districts and counties, This represents the rank-based average of the permanent resident population of all districts and counties. This represents the average of the official resident population statistics for all districts and counties.

[0094] (3) The Pearson correlation coefficient was used to assess the linear correlation between the unweighted and weighted permanent resident populations at the county / district level and the official statistical data. The calculation formula is as follows: .

[0095] .

[0096] in, The Pearson correlation coefficient is... For the first The number of permanent residents in each district / county, which can be either unweighted or weighted. For the first Official statistics on the permanent resident population of each district and county. d The number of districts and counties, This represents the average number of permanent residents across all districts and counties. This represents the average of the official resident population data for all districts and counties.

[0097] Step 4.3: Street and grid scale verification. Optionally, the grid cell resolution is 100m-500m.

[0098] (1) The Spearman correlation coefficient was used to calculate the ranking consistency between the unweighted resident population, the weighted resident population, and the aggregated value of the baseline population. The calculation formula is as follows: .

[0099] .

[0100] in, The Spearman correlation coefficient is... For the street / grid scale The number of permanent residents in a geographical unit, which can be either unweighted or weighted permanent residents; For the first Baseline population aggregation value for each geographic unit; , for , Rank of all geographic units at the street / grid scale; s This represents the total number of geographic units at the street / grid scale. This represents the rank average of the permanent resident population across all streets / grids. The average of the baseline population aggregation values ​​for all streets / grids.

[0101] (2) The linear correlation between the unweighted resident population, the weighted resident population, and the baseline population aggregate value at the street / grid scale was calculated using the Pearson correlation coefficient. The calculation formula is as follows: .

[0102] .

[0103] in, The Pearson correlation coefficient is... For the street / grid scale The number of permanent residents in a geographical unit, which can be either unweighted or weighted permanent residents; For the first The baseline population aggregation value for each geographic unit; s is the total number of geographic units at the street / grid scale. This represents the average number of permanent residents across all streets / grids. This is the average of the baseline population aggregation values ​​for all streets / grids.

[0104] Step 5: Multi-scale spatial model verification.

[0105] At multiple spatial scales (including at least street and grid scales), global spatial autocorrelation analysis and high-low clustering analysis are performed on the resident population identification result dataset to be verified and the high-resolution raster population dataset in the aggregated benchmark data to verify the consistency of their spatial patterns. Specifically, this includes: Specify the spatial scale.

[0106] Obtain three types of data for all geographic units at this spatial scale, including unweighted resident population, weighted resident population, and baseline population aggregation value.

[0107] Perform global spatial autocorrelation analysis on the three types of data respectively, calculate the Global Moran's I index and perform significance tests, as detailed in step 5.1.

[0108] Perform high-low value clustering analysis on the three types of data respectively, calculate the General G index and perform significance tests, and refer to step 5.2 for details.

[0109] Step 5.1: Global spatial autocorrelation analysis.

[0110] The Global Moran's I index is used to measure the overall spatial clustering or dispersion of population distribution. The calculation formula is as follows: .

[0111] .

[0112] in, For Global Moran's I Index; n This represents the total number of geographical units at the given spatial scale. For the first The attribute values ​​of each geographic unit (i.e., unweighted resident population, weighted resident population, or baseline population aggregate value). The elements of the spatial weight matrix represent geographic units. With geographical units Spatial proximity between (geographical units) With geographical units When adjacent to each other =1, otherwise =0, and =0); For the first The attribute values ​​of a geographic unit. This is the average of the attribute values ​​for all geographic units.

[0113] Global Moran's I value range and interpretation: I>0: positive spatial autocorrelation, similar attribute values ​​are clustered in space; I=0: no spatial autocorrelation, attribute values ​​are randomly distributed; I<0: negative spatial autocorrelation, dissimilar attribute values ​​are spatially close.

[0114] The significance of Z-scores is determined using the assumption of spatial random distribution. The formula for calculating the Z-score is as follows: .

[0115] in, Let I be the expected value. Let I be the variance of I, and n be the total number of geographic units at the given spatial scale. is the mathematical expectation of the square of I.

[0116] Look up the standard normal distribution table according to the Z-score to obtain the corresponding p-value.

[0117] The significance of spatial autocorrelation is judged according to the absolute value of the Z-score: if |Z|≥2.58 (corresponding to p≤0.01), it indicates that a very significant spatial autocorrelation exists; if 1.96≤|Z|<2.58 (corresponding to p≤0.05), it indicates that a significant spatial autocorrelation exists; if |Z|<1.96 (corresponding to p>0.05), it indicates no significance. When significant or very significant spatial autocorrelation exists, if I>E[I], it indicates positive spatial autocorrelation (the population presents an agglomeration distribution); if I<E[I], it indicates negative spatial autocorrelation (the population presents a dispersed distribution).

[0118] Step 5.2: High-low clustering analysis.

[0119] The General G index is used to identify whether high values (densely populated areas) or low values (sparsely populated areas) form significant clusters in space, and the calculation formula of the General G index is: .

[0120] Wherein, G is the General G index; n is the total number of geographic units at the spatial scale; , are respectively the attribute values of the th and the th geographic unit (that is, the unweighted permanent resident population number, the weighted permanent resident population number, or the reference population aggregation value); is the element of the spatial weight matrix, representing the spatial adjacency relationship between the geographic unit and the geographic unit (when the geographic unit and the geographic unit are adjacent, =1, otherwise =0, and =0).

[0121] The Z-score is used for significance judgment, and under the assumption of spatially random distribution, the calculation formula of the Z-score is: .

[0122] .

[0123] Wherein, is an element in the spatial weight matrix, representing the spatial adjacency relationship between the geographic unit and the geographic unit (the geographic unit and the geographic unit are adjacent to each other =1, otherwise =0, and =0); n is the total number of geographic units at said spatial scale, is the variance, is the mathematical expectation.

[0124] Look up the standard normal distribution table according to the Z-score to obtain the corresponding p-value.

[0125] The significance of high and low clustering is judged according to the absolute value of the Z-score: if |Z|≥2.58 (corresponding to p≤0.01), it indicates that a very significant high and low clustering exists; if 1.96≤|Z|<2.58 (corresponding to p≤0.05), it indicates that a significant high and low clustering exists; if |Z|<1.96 (corresponding to p>0.05), it indicates that the clustering is not significant. When significant or very significant high and low clustering exists, if G>E[G], it indicates high-value clustering ("hot spot", high population value areas are adjacent to each other), and if G<E[G], it indicates low-value clustering ("cold spot", low population value areas are adjacent to each other).

[0126] Step 6: Output of verification indicators.

[0127] All verification indicators calculated in Step 3 to Step 5 are systematically summarized and output. Said verification indicators include three categories, namely individual-level verification indicators, aggregation-level statistical verification indicators (or aggregation-level verification indicators), and multi-scale spatial pattern verification indicators.

[0128] Individual-level verification indicator: the recognition accuracy of the permanent resident population based on high-confidence overlapping samples (wherein the recognition accuracy is a conservative lower bound estimate).

[0129] Aggregation-level statistical verification indicators (or aggregation-level verification indicators): urban-level absolute and relative errors of the unweighted permanent resident population and the weighted permanent resident population; district and county-level absolute and relative errors of the weighted permanent resident population; district and county-level Spearman correlation coefficient and Pearson correlation coefficient; subdistrict-level Spearman correlation coefficient and Pearson correlation coefficient; grid-level Spearman correlation coefficient and Pearson correlation coefficient.

[0130] Multi-scale spatial pattern verification indicators: the Global Moran's I index and its Z-score and p-value, and the General G index and its Z-score and p-value of the unweighted permanent resident population, the weighted permanent resident population, and the aggregated reference population value at each scale (subdistrict, grid).

[0131] Optionally, the Z-score and p-value of the Global Moran's I index and the General G index are calculated through Monte Carlo simulation, for judging the significance of spatial agglomeration or clustering.

[0132] The above indicators are output in the form of standardized numerical tables and statistical charts (including but not limited to scatter plots, spatial clustering diagrams, and indicator summary tables), serving as the core basis for quantitatively assessing the reliability of the resident population identification results.

[0133] Compared with related technologies, this application has the following significant advantages: 1. Pioneering an individual-level verification system based on conservative benchmarks: By constructing high-confidence positive samples with the overlap of operators' conservative "family location" data and the results to be verified, a large-scale individual-level accuracy lower bound estimation of the resident population identification results is achieved, filling the gap in individual-level verification in related technologies and solving the technical problem of unreliable individual-level benchmark data.

[0134] 2. Comprehensive Spatial Validation Dimensions: Based on traditional numerical correlation validation, Global Moran's I global spatial autocorrelation analysis and General G high-low clustering analysis are introduced to fully cover the dual validation requirements of "numerical consistency - spatial pattern consistency" and overcome the shortcomings of the single spatial validation dimension of related technologies.

[0135] 3. Construct a multi-scale verification framework: Clearly define a full-scale verification system covering individuals, grids, streets, districts / counties, and cities. This system can accurately assess the reliability of identification results at different spatial granularities and meet the application needs of multiple scenarios such as urban planning and population statistics.

[0136] 4. Optimize the expansion and statistical process: Adopt a two-stage spatial attribution strategy that combines statistical expansion weights with "Geohash coarse filtering + geofencing fine filtering", which not only ensures the accuracy of the aggregation-level estimation, but also improves the processing efficiency of large-scale data, and solves the problems of the lack of rigor in the expansion logic and low statistical efficiency of related methods.

[0137] 5. Output a standardized quantitative indicator set: All verification indicators are quantifiable objective indicators, which facilitates horizontal comparison between different permanent resident identification algorithms and provides a clear quantitative direction for algorithm optimization.

[0138] This embodiment takes City XA as an example to provide a detailed description of the verification method for the permanent resident identification results of this application.

[0139] Step 1: Obtain the multi-source benchmark dataset and the dataset of resident population identification results to be verified.

[0140] Step 1.1, Dataset of Resident Identification Results to be Verified: Based on mobile phone signaling data of City XA from January to June 2022 and July 2023, the dataset of resident identification using the spatiotemporal stability identification framework contains 3,180,905 users; the attribute information of each user record is as follows: User unique identifier: Anonymous user ID assigned by the operator.

[0141] Residential location information: latitude and longitude coordinates Level 6 Geohash grid coding ( ).

[0142] Statistical expansion weighting ( : A unique coefficient assigned by the telecommunications operator to each user, used to characterize the user's representativeness to the overall population.

[0143] Step 1.2, Multi-source benchmark datasets: Individual-level conservative benchmark data: "Home location" data of XA City users provided by China Unicom. This data records the location with the longest cumulative nighttime stay nationwide since the SIM card was activated. The update rule is "the update is triggered when the cumulative nighttime stay of the new location exceeds that of the original location", reflecting the user's long-term historical stable residential location; Aggregate-level benchmark data: XA City Statistics Bureau's 2022 year-end resident population statistics (2022XRP) is 12,995,900 people; the total number of market users (conU) of telecommunications operators in XA City is 3,183,628 people; WorldPop 2020 100-meter resolution raster population dataset.

[0144] Step 2: Individual-level verification.

[0145] Step 2.1, User Matching: The 3,180,905 permanent residents to be verified are matched with the individual-level conservative baseline data using unique user identifiers to obtain overlapping user sets. It contains 2,899,017 users; the residential location stability of this overlapping user set has been verified by the operator's long-term nighttime stay behavior, and is a high-confidence positive sample.

[0146] Step 2.2, Accuracy Calculation: According to the formula Substituting the data, we get: 2,899,017 / 3,180,905=0.9114, which means the individual-level recognition accuracy is 91.14%. This value is a conservative lower bound estimate, and the actual recognition accuracy is not lower than this value.

[0147] Step 3: Aggregate-level statistical validation.

[0148] Step 3.1: Execution of the two-stage statistical process.

[0149] Coarse filtering stage: based on 6-level Geohash grid coding ( Construct a spatial index and filter out each geographic unit (city, district / county, street, 200m grid) separately. For potential users whose data intersects with the unit, the average data reduction rate is 85%.

[0150] Fine filtering stage: Geofencing is performed on the latitude and longitude coordinates of potential users to determine the attribution user set for each unit.

[0151] Weighted summation: For each unit's attribution user set The numbers are summed to obtain the weighted resident population.

[0152] Step 3.2, Scaled verification results.

[0153] Urban-scale verification: (1) The unweighted permanent resident population = 3,180,905 (total number of users attributable to the user base), and the absolute relative error between the total number of users in the operator market (3,183,628) is |3,180,905-3,183,628| / 3,183,628×100%≈0.1%.

[0154] (2) Weighted permanent resident population (All users) The sum of the values ​​is 13,220,833, and the absolute relative error between the sum and the official statistics (12,995,900) is |13,220,833-12,995,900| / 12,995,900×100%≈1.7%.

[0155] District / county level verification: (1) The Spearman correlation coefficient between the unweighted permanent resident population and the 2022 XRP was 0.986, and the Pearson correlation coefficient was 0.959.

[0156] (2) The Spearman correlation coefficient between the weighted permanent resident population and the 2022 XRP was 0.909, and the Pearson correlation coefficient was 0.936.

[0157] Street-scale verification: (1) The Spearman correlation coefficient between the unweighted permanent resident population and the WorldPop aggregate value was 0.814, and the Pearson correlation coefficient was 0.859.

[0158] (2) The Spearman correlation coefficient between the weighted permanent resident population and the WorldPop aggregate value is 0.815, and the Pearson correlation coefficient is 0.834.

[0159] Grid scale verification (200m): (1) The Spearman correlation coefficient between the unweighted permanent resident population and the WorldPop aggregate value was 0.651, and the Pearson correlation coefficient was 0.567.

[0160] (2) The Spearman correlation coefficient between the weighted permanent resident population and the WorldPop aggregate value was 0.654, and the Pearson correlation coefficient was 0.565.

[0161] Step 4: Multi-scale spatial model verification.

[0162] Step 4.1, Verification of street-scale spatial patterns.

[0163] Global spatial autocorrelation analysis: The Global Moran's I index for the unweighted permanent resident population in the dataset to be verified is 0.5164 (Z=12.02, p=0.0000), the Global Moran's I index for the weighted permanent resident population is 0.5053 (Z=11.89, p=0.0000), and the Global Moran's I index for the WorldPop aggregation value is 0.4498 (Z=10.37, p=0.0000). All three show significant spatial clustering patterns.

[0164] High-low clustering analysis: The General G index of the unweighted permanent resident population in the dataset of the resident population identification results to be verified is 0.012290 (expected value = 0.00510, Z = 11.92, p = 0.0000), the General G index of the weighted permanent resident population is 0.01186 (expected value = 0.00510, Z = 11.76, p = 0.0000), and the General G index of the WorldPop aggregation value is 0.00730 (expected value = 0.00510, Z = 10.34, p = 0.0000). All three show a significant high-value clustering pattern.

[0165] Step 4.2, Grid-scale (200m) spatial pattern verification.

[0166] Global spatial autocorrelation analysis: The Global Moran's I index for the unweighted permanent resident population in the dataset of permanent resident identification results to be verified is 0.603 (Z=604.91, p=0.0000), the Global Moran's I index for the weighted permanent resident population is 0.6132 (Z=614.75, p=0.0000), and the Global Moran's I index for the WorldPop aggregate value is 0.8901 (Z=892.00, p=0.0000). Both show a strong and significant spatial clustering pattern.

[0167] High-low clustering analysis: The General G index of the unweighted permanent resident population in the dataset to be verified is 0.000083 (expected value = 0.000004, Z = 604.81, p = 0.0000), the General G index of the weighted permanent resident population is 0.000083 (expected value = 0.000004, Z = 614.64, p = 0.0000), and the General G index of the WorldPop aggregation value is 0.000024 (expected value = 0.000004, Z = 891.15, p = 0.0000). Both show significant high-value clustering patterns.

[0168] Step 5: Verify the output of the indicators.

[0169] A standardized verification report is generated by summarizing all the above verification indicators. The core indicators are shown in Table 1. Table 1 Summary of Verification Indicators

[0170] The above indicators are output in the form of tables, scatter plots and spatial clustering diagrams, providing a quantitative assessment basis for the reliability of this resident population identification method.

[0171] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 2 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores verification data for resident identification results. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for verifying resident identification results.

[0172] Those skilled in the art will understand that Figure 2The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0173] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0174] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0175] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0176] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0177] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0178] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0179] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0180] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for verifying the identification results of permanent residents, characterized in that, include: Acquire the dataset of permanent resident identification results to be verified based on mobile signaling data, individual-level conservative benchmark data, and aggregate-level benchmark data; The resident population identification result dataset includes user sets and attribute information of user sets; the aggregated benchmark data includes official statistical resident population data, total number of users in the operator market, and high-resolution raster population dataset; wherein, the high resolution range is less than or equal to 500m; The unweighted and weighted permanent resident population numbers are calculated at multiple spatial scales based on the dataset of permanent resident population identification results to be verified; the spatial scales include: city scale, district / county scale, street scale, and grid scale. Calculate baseline population aggregation values ​​across multiple spatial scales based on high-resolution raster population datasets; The dataset of permanent resident identification results to be verified is matched with individual-level conservative benchmark data using user unique identifiers to calculate individual-level verification indicators. At different spatial scales, aggregate-level verification indicators are calculated for each spatial scale based on unweighted permanent resident population, weighted permanent resident population, and aggregate-level baseline data. Global spatial autocorrelation and cluster analysis were performed on the unweighted permanent resident population, weighted permanent resident population, and baseline population aggregation value at the street and grid scales, respectively, to obtain multi-scale spatial pattern verification indicators. Individual-level verification metrics, aggregate-level verification metrics, and multi-scale spatial model verification metrics are summarized.

2. The method for verifying the identification results of permanent residents according to claim 1, characterized in that, Based on the dataset of resident population identification results to be verified, calculate the unweighted and weighted resident population numbers at multiple spatial scales, specifically including: For each geographic unit at multiple spatial scales, a two-stage spatial filtering strategy of coarse filtering and fine filtering is used to determine the residential location of users in the dataset of permanent residents to be verified to belong to the attribution user set of each geographic unit. The unweighted resident population of each geographic unit is obtained from the length of the attribution user set; The weighted resident population of each geographic unit is obtained by summing the statistical expansion weights of all users in the attribution user set.

3. The method for verifying the identification results of permanent residents according to claim 2, characterized in that, For each geographic unit across multiple spatial scales, a two-stage spatial filtering strategy involving coarse and fine filtering is employed to determine the attribution user set of each geographic unit within the resident population identification result dataset to be validated. Specifically, this includes: For each geographic unit at multiple spatial scales, calculate the Geohash encoding set covered by the geographic unit; Based on the latitude and longitude coordinates of the user's residential location in the dataset of permanent resident identification results to be verified, calculate the Geohash code corresponding to the latitude and longitude coordinates of the user's residential location. The candidate user set is obtained by filtering users whose Geohash codes corresponding to the latitude and longitude coordinates of their residential location belong to the set of Geohash codes covered by the geographic unit; Based on the candidate user set, the latitude and longitude coordinates of the candidate users' residential locations are analyzed with geographic units using GIS spatial intersection analysis to determine the user set whose residential location is within the geographic unit, which is then used as the attribution user set.

4. The method for verifying the identification results of permanent residents according to claim 1, characterized in that, Calculate baseline population aggregation values ​​across multiple spatial scales based on high-resolution raster population datasets, including: For each geographic unit at multiple spatial scales, determine the set of raster cells covered by each geographic unit in the high-resolution raster population dataset; Extract the population value of each raster cell in the raster cell set and sum them to obtain the baseline population aggregate value of each geographic unit.

5. The method for verifying the identification results of permanent residents according to claim 1, characterized in that, The individual-level verification metric includes recognition accuracy, and the formula for calculating recognition accuracy is as follows: ； in, To improve recognition accuracy, This refers to the number of users in the overlapping user set obtained from the dataset of resident population identification results to be verified and the individual-level conservative benchmark data. This represents the total number of permanent residents in the dataset of permanent resident identification results to be verified.

6. The method for verifying the identification results of permanent residents according to claim 1, characterized in that, The aggregated verification indicators for each spatial scale include: the absolute relative error of the unweighted permanent resident population at the city scale, the absolute relative error of the weighted permanent resident population at the city scale, the absolute relative error of the weighted permanent resident population at the district / county scale, the Spearman correlation coefficient of the unweighted permanent resident population at the district / county scale, the Spearman correlation coefficient of the weighted permanent resident population at the district / county scale, the Pearson correlation coefficient of the unweighted permanent resident population at the district / county scale, the Pearson correlation coefficient of the weighted permanent resident population at the district / county scale, the Spearman correlation coefficient of the weighted permanent resident population at the street scale, the Pearson correlation coefficient of the weighted permanent resident population at the street scale, the Spearman correlation coefficient of the weighted permanent resident population at the grid scale, and the Pearson correlation coefficient of the weighted permanent resident population at the grid scale.

7. The method for verifying the identification results of permanent residents according to claim 1, characterized in that, The multi-scale spatial model validation indices include: the Global Moran's I index of the unweighted resident population of the street-scale spatial model, the Global Moran's I index of the weighted resident population of the street-scale spatial model, the Global Moran's I index of the baseline population aggregation value of the street-scale spatial model, the General G index of the unweighted resident population of the street-scale spatial model, the General G index of the weighted resident population of the street-scale spatial model, the General G index of the baseline population aggregation value of the street-scale spatial model, the Global Moran's I index of the unweighted resident population of the grid-scale spatial model, the Global Moran's I index of the weighted resident population of the grid-scale spatial model, the Global Moran's I index of the baseline population aggregation value of the grid-scale spatial model, the General G index of the unweighted resident population of the grid-scale spatial model, the General G index of the weighted resident population of the grid-scale spatial model, and the General G index of the baseline population aggregation value of the grid-scale spatial model.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the method for verifying the resident population identification results according to any one of claims 1-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the verification method for the identification results of permanent residents as described in any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the verification method for the identification results of permanent residents as described in any one of claims 1-7.

Images