Business circle type division method and device and electronic equipment

Abstract

The application provides a business circle type division method and device and electronic equipment, and belongs to the technical field of data analysis. The method comprises the following steps: for each POI clustering unit corresponding to a to-be-classified business circle, determining a user use feature of the POI clustering unit according to a function behavior category and a time behavior category corresponding to a plurality of effective user visit behavior records of the POI clustering unit in a statistical period; determining a multi-dimensional classification feature vector of the POI clustering unit according to the user use feature and a preset classification feature of the POI clustering unit; and clustering the multi-dimensional classification feature vectors of the POI clustering units to obtain a business circle type division result of the to-be-classified business circle corresponding to each POI clustering unit. According to the application, the effective user visit behavior records representing user visit behaviors are introduced for business circle classification on the basis of static commercial service POIs, so that the description ability of the business circle type division result to the real use condition and service capacity of the business circle can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of data analysis technology, and in particular to a method, apparatus, and electronic device for classifying business districts. Background Technology

[0002] Scientific and stable identification and classification of business districts is an important prerequisite for understanding urban commercial structure, assessing differences in commercial vitality, and formulating refined management policies.

[0003] Common business district classification schemes are based on the commercial supply structure. This involves identifying the spatial scope of the business district and then, based on the supply structure of commercial facilities or the composition of business formats, counting the number or proportion of different types of points of interest (POIs) within the business district to classify the business district into functional types.

[0004] However, in practical applications, different business districts may exhibit significant differences in visit frequency, dwell time, usage time distribution, and customer demographics. Even with similar commercial facility supply structures, their actual service capabilities and consumption functions may vary considerably. Therefore, business district classification schemes based on commercial supply structures that rely solely on static POI distribution characteristics for functional classification cannot fully depict the true usage and service capabilities of business districts, thus limiting the explanatory power of the classification results in refined analysis and practical applications. Summary of the Invention

[0005] This invention provides a method, apparatus, and electronic device for classifying business district types, which addresses the shortcomings of existing business district classification schemes based on commercial supply structure that rely solely on static POI distribution characteristics for functional classification. These schemes fail to fully depict the actual usage and service capabilities of business districts, thus limiting the explanatory power of classification results in refined analysis and practical applications.

[0006] This invention provides a method for classifying business district types, including: Spatial clustering is performed on commercial service points (POIs) in target areas that require business district classification, resulting in multiple POI clustering units; wherein, each POI clustering unit corresponds to a business district to be classified. For each POI clustering unit, the user usage characteristics of the POI clustering unit are determined based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within the statistical period; and the multidimensional classification feature vector of the POI clustering unit is determined based on the user usage characteristics and the preset classification characteristics of the POI clustering unit. Clustering is performed on the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit; The valid user visit behavior records are determined based on user equipment signaling residency records whose visit times match the business hours of the business service POI.

[0007] According to a business district type classification method provided by the present invention, the step of determining the user usage characteristics of the POI clustering unit based on the functional behavior categories and time behavior categories corresponding to multiple valid user visit behavior records of the POI clustering unit within a statistical period includes: determining the functional behavior category proportion vector of the POI clustering unit based on the number of valid user visit behavior records for each functional behavior category; determining the time behavior category proportion vector of the POI clustering unit based on the number of valid user visit behavior records for each time behavior category; obtaining the functional diversity characteristics of the POI clustering unit based on the information entropy of multiple valid user visit behavior records; and determining the user usage characteristics of the POI clustering unit based on the functional behavior category proportion vector, the time behavior category proportion vector, and the functional diversity characteristics.

[0008] According to a business district classification method provided by the present invention, the functional behavior categories corresponding to multiple valid user visit records are determined based on the following method: For each valid user visit record, at least two spatial buffers of different areas are delineated with the visit location in the valid user visit record as the geometric center of the spatial buffer; the business service POI type proportion vectors corresponding to each spatial buffer are weighted and summed according to preset weights to obtain the event-level functional exposure vector of the valid user visit record; the event-level functional exposure vectors of each valid user visit record are clustered to obtain the functional behavior categories corresponding to each valid user visit record; wherein, the preset weights are inversely correlated with the area of ​​the spatial buffer; the business service POI type proportion vector corresponding to the spatial buffer is determined according to the proportion of the number of business service POIs of different business service types within the spatial buffer.

[0009] According to a business district type division method provided by the present invention, the step of delineating at least two spatial buffers of different areas with the visited locations in the effective user visit behavior records as the geometric centers of the spatial buffers, and weighting and summing the business service POI type proportion vectors corresponding to each of the spatial buffers according to preset weights, includes: delineating a first spatial buffer, a second spatial buffer, and a third spatial buffer with the visited locations in the effective user visit behavior records as the geometric centers of each of the spatial buffers; the area of ​​the first spatial buffer is smaller than the area of ​​the second spatial buffer, and the area of ​​the second spatial buffer is smaller than the area of ​​the third spatial buffer; weighting and summing the first business service POI type proportion vector corresponding to the first spatial buffer, the second business service POI type proportion vector corresponding to the second spatial buffer, and the third business service POI type proportion vector corresponding to the third spatial buffer according to a first weight, a second weight, and a third weight, respectively; the first weight is greater than the second weight, and the second weight is greater than the third weight.

[0010] According to a business district classification method provided by the present invention, the preset classification features include the customer source ratio feature of the POI cluster unit; the customer source ratio feature of the POI cluster unit is determined according to the following method: the number of residential customer source users of the POI cluster unit is determined according to the distance between the geographical location of the business district to be classified corresponding to the POI cluster unit and the residential location of the effective user visit behavior record within the POI cluster unit; the number of workplace customer source users of the POI cluster unit is determined according to the distance between the geographical location of the business district to be classified corresponding to the POI cluster unit and the workplace location of the effective user visit behavior record within the POI cluster unit; the residential customer source ratio and the workplace customer source ratio are determined according to the number of residential customer source users and the number of workplace customer source users; the customer source ratio feature of the POI cluster unit is determined according to the residential customer source ratio and the workplace customer source ratio.

[0011] According to a business district classification method provided by the present invention, the preset classification features further include behavioral intensity features, dwell depth features, and spatial structure features of the POI clustering units; the behavioral intensity features are determined based on the scale of visits, the percentage of daily peak visits, and the frequency of visits per person; the dwell depth features are determined based on the average dwell time of users and the percentage of users who spend a long time; and the spatial structure features are determined based on the maximum value of POI kernel density.

[0012] According to a business district classification method provided by the present invention, the step of spatially clustering commercial service POIs in the target area that needs to be classified into business districts to identify multiple POI clustering units includes: spatially clustering the commercial service POIs according to a first radius to obtain local high-density POI subclusters; spatially clustering the local high-density POI subclusters according to a second radius to obtain POI clustering units; wherein the second radius is greater than the first radius.

[0013] According to a method for classifying business districts provided by the present invention, the first radius is determined based on the building radius of the target area; the second radius is determined based on the street radius of the target area.

[0014] The present invention also provides a business district type classification device, comprising: The business district identification module is used to spatially cluster commercial service points (POIs) in target areas that need to be classified into business districts, and to identify multiple POI clustering units; wherein, each POI clustering unit corresponds to a business district to be classified. The multidimensional vector acquisition module is used to determine the user usage characteristics of each POI clustering unit based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within the statistical period; and to determine the multidimensional classification feature vector of the POI clustering unit based on the user usage characteristics and the preset classification characteristics of the POI clustering unit. The business district classification module is used to cluster the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit; The valid user visit behavior records are determined based on user equipment signaling residency records whose visit times match the business hours of the business service POI.

[0015] The present invention also provides an electronic 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 any of the above-described business district type division methods.

[0016] The present invention provides a method, apparatus, and electronic device for classifying business districts. Based on static commercial service point of interest (POI) data, it introduces effective user visit behavior records that characterize actual user visits and usage behavior. Utilizing the functional and temporal behavior categories of these records, it determines the user usage characteristics of POI clustering units corresponding to the business district to be classified. Then, it constructs multidimensional classification feature vectors for each POI clustering unit using these user usage characteristics and preset classification features. After clustering the multidimensional classification feature vectors of each POI clustering unit, it obtains the business district classification results. This enables urban business district classification analysis based on user visit behavior, effectively improving the ability of the business district classification results to depict the actual usage and service capabilities of the business districts, and enhancing the explanatory power of the classification results in refined analysis and practical applications. Attached Figure Description

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

[0018] Figure 1 This is one of the flowcharts illustrating the business district type classification method provided by the present invention.

[0019] Figure 2 This is the second flowchart of the business district type classification method provided by the present invention.

[0020] Figure 3 This is an example diagram of the local high-density POI sub-clusters and POI clustering units provided by the present invention.

[0021] Figure 4 This is a schematic diagram of the business district type classification device provided by the present invention.

[0022] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0024] The following is combined with Figures 1 to 5This invention describes a business district type classification method, apparatus, and electronic device.

[0025] A commercial district, also known as a business cluster, is a spatial unit in a city where commercial facilities and consumer activities are highly concentrated. It is an important component of the urban commercial system and plays a fundamental role in urban planning, commercial layout optimization, public service allocation, commercial operation management, and consumer behavior research. The scientific and stable identification and classification of commercial districts is a crucial prerequisite for understanding the urban commercial structure, assessing differences in commercial vitality, and formulating refined management policies.

[0026] Business district identification is the process of determining the spatial extent of a business district within an urban space.

[0027] Common business district identification technologies are all based on a single scale, specifically including those based on regular spatial units and those based on road network street block division. The regular spatial unit-based approach typically identifies commercial clusters using regularized spatial units (such as grids or lattices). It divides the study area into fixed-size regular grid units, counts the number or density of commercial service points of interest (POIs) within each grid, and identifies relatively concentrated areas of commercial activity based on threshold filtering or clustering algorithms. The road network street block-based approach considers the road network as the basic framework of urban spatial structure, and street blocks as a more accurate analytical unit reflecting the actual spatial form of the city. Therefore, it generally uses the urban road network as a foundation, dividing urban space into street block units and analyzing commercial vitality or density at the street block scale.

[0028] Business district classification is based on business district identification, and it involves dividing different business districts according to their functional structure or type.

[0029] Common business district classification schemes are based on the commercial supply structure. This involves identifying the spatial scope of the business district and then, based on the supply structure of commercial facilities or the composition of business formats, counting the number or proportion of different types of Points of Interest (POIs) within the business district to classify the business district into functional types.

[0030] However, in practical applications, different business districts may exhibit significant differences in visit frequency, dwell time, usage time distribution, and customer demographics. Even with similar commercial facility supply structures, their actual service capabilities and consumption functions may vary considerably. Therefore, business district classification schemes based on commercial supply structures that rely solely on static POI distribution characteristics for functional classification cannot fully depict the true usage and service capabilities of business districts, thus limiting the explanatory power of the classification results in refined analysis and practical applications.

[0031] In view of this, the present invention provides a method, apparatus and electronic device for classifying business districts to solve the aforementioned problems.

[0032] It should be noted that all actions of acquiring signals, information or data in the business district type classification method, device and electronic equipment provided by the present invention are carried out in compliance with the relevant data protection laws and policies of the local country and with authorization from the owner of the corresponding device.

[0033] Figure 1 This is one of the flowcharts illustrating the business district type classification method provided by the present invention, such as... Figure 1 As shown, the business district type classification method includes, but is not limited to, steps 101 to 103.

[0034] It should be noted that the execution subject of the business district type classification method provided by the present invention can be a server, computer equipment, such as mobile phone, tablet computer, laptop computer, handheld computer, vehicle electronic equipment, wearable device, ultra-mobile personal computer (UMPC), netbook or personal digital assistant (PDA), etc.

[0035] Step 101: Spatial clustering of commercial service POIs in the target area that needs to be classified into business districts is performed to identify multiple POI clustering units.

[0036] In this context, each POI clustering unit corresponds to a business district to be classified.

[0037] A Point of Interest (POI) for business services is a data unit in a geographic information system used to mark the location of a business service location and to represent the entity of a business service location.

[0038] The types of Business Service Points of Interest (POIs) include, but are not limited to, Food and Beverage Service POIs, Shopping Service POIs, Lifestyle Service POIs, and Sports and Leisure Service POIs. Specifically, Food and Beverage Service POIs can include restaurants, tea houses, bakeries, cafes, fast food restaurants, beverage shops, dessert shops, foreign restaurants, casual dining venues, and Chinese restaurants, etc. Shopping Service POIs can include convenience stores, supermarkets, clothing, footwear and leather goods stores, personal care / cosmetics stores, shopping-related venues, flower, bird, fish and insect markets, home appliance and electronics stores, home building materials markets, shopping malls, specialty commercial streets, sporting goods stores, stationery stores, specialty stores, and general markets. Lifestyle Service POIs can include beauty salons, photo printing shops, lifestyle service venues, laundromats, and bath and massage parlors. Sports and Leisure Service POIs can include sports and leisure service venues, leisure venues, cinemas, entertainment venues, and sports venues.

[0039] Specifically, Figure 2 This is the second flowchart of the business district type classification method provided by the present invention, combined with... Figure 1 and Figure 2 As shown, based on POI data provided by map software, various commercial service POIs highly related to urban commercial activity clusters (i.e., business districts) within a predetermined target area (such as a city) that needs to be classified into business districts are selected. Spatial clustering methods are used to spatially cluster the numerous commercial service POIs within the business district to be classified, identifying multiple POI clustering units, where each POI clustering unit corresponds to a business district to be classified.

[0040] Optionally, the specific clustering method for spatial clustering of commercial service POIs in the target area that needs to be classified into business districts is the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) method.

[0041] Step 102: For each POI clustering unit, determine the user usage characteristics of the POI clustering unit based on the functional behavior category and time behavior category corresponding to the multiple valid user visit behavior records of the POI clustering unit within the statistical period; determine the multidimensional classification feature vector of the POI clustering unit based on the user usage characteristics and the preset classification characteristics of the POI clustering unit.

[0042] Among them, valid user visit behavior records are determined based on user equipment signaling residency records whose visit times conform to the business hours of the commercial service POI. User equipment signaling residency records are detailed log data records used to indicate the dwell status of users and their carried user equipment in the spatial grid defined by the telecommunications operator. If the visit time in the user equipment signaling residency record does not conform to the business hours of the commercial service POI, then the user equipment signaling residency record cannot be used to determine valid user visit behavior records.

[0043] Based on user equipment signaling residency records, event-level structured features of user visit behavior can be extracted to obtain valid user visit behavior records, including but not limited to start time stime, end time etime, date type date_type (used to distinguish between weekdays and rest days), latitude and longitude of stay (lat, lon), cluster ID (also known as POI clustering unit ID), and the user's residential location and work location (home_lat, home_lon, work_lat, work_lon).

[0044] By extracting event-level structured features of user visit behavior from user equipment signaling residency records, it is possible to transform the raw signaling trajectory into standardized single visit behavior samples, providing a data foundation for subsequent functional exposure analysis, time behavior clustering, and the construction of a business district classification indicator system.

[0045] The visit time includes the start and end times of the user's stay as recorded in the user equipment signaling residency log. For the business hours of commercial service POIs, specific settings can be made based on the typical business hours of commercial complexes and various shops in the city's business district. The business hours of commercial service POIs of different categories, locations, and entities can be the same or different; there are no restrictions on this.

[0046] For example, valid user visit records are user equipment signaling records that meet the filtering criteria of "start time stime > 8:00 and end time etime < 23:50", where "8:00~23:50" is the set business hours.

[0047] The functional behavior categories corresponding to valid user visit records are classification tags based on the commercial service functions associated with user visit behavior, used to describe their functional attributes in commercial services. For example, functional behavior categories include, but are not limited to, dining behavior, shopping behavior, entertainment behavior, lifestyle service behavior, and comprehensive service behavior.

[0048] The time-based behavior categories corresponding to valid user visit records are classification tags based on the time type associated with the user's visit behavior. For example, time-based behavior categories include, but are not limited to, weekday noon behavior, weekday evening behavior, weekend noon behavior, weekend evening behavior, and weekend afternoon behavior.

[0049] Preset classification features are characteristics of the POI clustering unit that are determined according to a pre-defined business district classification index system, other than user usage characteristics. Preset classification features include, but are not limited to, at least one of the following features: behavioral intensity features, dwell time features, customer source features, and spatial structure features.

[0050] To improve the ability of business district classification results to accurately depict the actual usage and service capabilities of business districts and enhance the explanatory power of the classification results, this invention introduces user actual visit and usage behavior information on the basis of static POI data, in order to provide a city business district classification method based on user visit behavior analysis.

[0051] Specifically, in combination Figure 2 As shown, after identifying multiple POI clustering units in the target area requiring business district classification, the multiple POI clustering units undergo business district grid mapping processing, including mapping each POI clustering unit to a spatial grid of a preset size (e.g., 150 meters) provided by the telecom operator. Here, relying on the user equipment signaling registration records and user grid provided by the telecom operator, the spatial grid can be directly used as a statistical unit for user visit behavior events.

[0052] By leveraging the DaaS (Desktop as a Service) platform of telecom operators, we obtain all or part of the user equipment signaling residency records within all spatial grids corresponding to all POI clustering units within a preset statistical period (e.g., December). Combined with the business hours of commercial service POIs, we extract event-level structured features of user visit behavior from user equipment signaling residency records whose visit times match the business hours of commercial service POIs, thus obtaining valid user visit behavior records.

[0053] Each valid user visit record has a corresponding functional behavior category and time behavior category to describe the user's visit behavior. Therefore, according to the predetermined functional behavior analysis method and time behavior analysis method, the functional behavior category and time behavior category corresponding to each of the valid user visit records are determined.

[0054] The key to classifying business districts by incorporating actual user visits and usage behavior information lies in constructing a multi-dimensional classification feature vector for each POI clustering unit based on valid user visit behavior records.

[0055] For each POI cluster unit, the user usage characteristics of the POI cluster unit are determined by using the functional behavior category and time behavior category corresponding to all valid user visit records of the POI cluster unit within the statistical period, according to the preset user usage characteristic determination method. Furthermore, the user usage characteristics of the POI cluster unit, as well as the preset classification characteristics of the POI cluster unit determined according to the preset business district classification index system, are used to determine the multidimensional classification feature vector of the POI cluster unit.

[0056] By repeating the steps of determining the user usage characteristics of POI cluster units based on the functional behavior categories and time behavior categories corresponding to multiple valid user visit behavior records of POI cluster units within the statistical period, and determining the multidimensional classification feature vector of POI cluster units based on the user usage characteristics and the preset classification characteristics of POI cluster units, the multidimensional classification feature vector of all POI cluster units in the commercial area to be classified can be obtained.

[0057] Optionally, after performing business district grid mapping on multiple POI clustering units, grids with a number of commercial service POIs greater than or equal to a preset threshold (e.g., 5, 10) within the spatial grid are identified as valid spatial grids, and invalid spatial grids with a number of commercial service POIs less than the preset threshold are deleted to eliminate interference from isolated points and low-density areas, ensuring that POI clustering units have a certain commercial agglomeration intensity and a stable commercial foundation.

[0058] Step 103: Cluster the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit.

[0059] Specifically, the K-means clustering method is employed, using the multidimensional classification feature vectors of each POI cluster unit as input. By minimizing the squared error between samples within a cluster and their corresponding cluster centers, the multidimensional classification feature vectors of all POI cluster units within the target area requiring business district classification are clustered, achieving automatic grouping of business districts in the feature space. After the clustering calculation is completed, based on the differences in the values ​​of each cluster center across different behavioral and spatial feature dimensions, the feature structures of different categories of business districts are interpreted. This identifies typical characteristics of various business districts in terms of visitor scale, dwell time, functional structure, temporal rhythm, and spatial agglomeration, ultimately resulting in a business district type classification with clear meaning, corresponding to each POI cluster unit.

[0060] After obtaining the business district type classification results corresponding to the POI clustering units, the business district type classification results can be used in application scenarios such as consumer behavior research and user usage pattern analysis. They can also be used in application scenarios such as urban business district planning and commercial space layout optimization that require coverage of a large area and use of full user data. Furthermore, they can be used in application scenarios such as commercial vitality monitoring, operation evaluation, and trend analysis that require characterizing the dynamic changes of business districts.

[0061] Based on the classification results of business district types, the business districts to be classified in the POI clustering unit can be divided into any one of the following: city-level multi-functional integrated business district, regional integrated service business district, mixed-use daily consumption business district, pure shopping destination business district, and leisure experience long-stay business district, to support applications such as urban business district classification, functional positioning optimization, and commercial facility layout planning.

[0062] Optionally, clustering is performed on the multidimensional classification feature vectors of each POI clustering unit within the target area requiring business district classification, according to the first cluster number. The first cluster number is determined based on the scree plotting principle. A scree plot is drawn by calculating the sum of squares within each cluster under different cluster numbers, and a reasonable number of clusters is determined based on the inflection point where the curve transitions from steep to gentle, balancing the stability and interpretability of the classification results. In other words, the first cluster number is determined based on the sum of squares within each multidimensional classification feature vector under different cluster numbers.

[0063] The business district classification method provided by this invention introduces effective user visit behavior records that characterize actual user visits and usage behavior information on the basis of static commercial service POI data. Using the functional and temporal behavior categories of these records, the method determines the user usage characteristics of POI clustering units corresponding to the business district to be classified. Then, it constructs multi-dimensional classification feature vectors for each POI clustering unit using these user usage characteristics and preset classification features. After clustering the multi-dimensional classification feature vectors of each POI clustering unit, the business district classification result is obtained. This enables urban business district classification analysis based on user visit behavior, effectively improving the ability of the business district classification results to depict the actual usage and service capabilities of the business districts, and enhancing the explanatory power of the classification results in refined analysis and practical applications.

[0064] Furthermore, compared to methods for classifying business districts based on user profiles, the business district type classification method provided by this invention transforms the analysis object from a single user profile corresponding to a single user into multiple user device signaling residency records that can represent multiple visit events for a single user. These user device signaling residency records are then filtered and structured to extract start and end times, dwell time, midpoint time, date type, dwell coordinates, and work / residence information for each visit event. Effective user visit behavior records that can be clustered are then constructed on an event-by-event basis. This overcomes the shortcomings of user profiles in business district research, such as "sparse visits and unstable feature aggregation," and avoids the problem of unclear behavioral patterns caused by sparse user visits. It ensures that behavioral characteristics reflect real consumption patterns and provides a high-quality sample foundation for functional identification and temporal rhythm analysis.

[0065] Based on the above embodiments, as an optional embodiment, determining the user usage characteristics of the POI clustering unit according to the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within the statistical period includes: The functional behavior category proportion vector of the POI clustering unit is determined based on the number of valid user visit records for each functional behavior category. Based on the number of valid user visit records for each time behavior category, determine the time behavior category proportion vector of the POI clustering unit; Based on the information entropy of multiple valid user visit records, the functional diversity characteristics of the POI clustering unit are obtained; Based on the functional behavior category proportion vector, the time behavior category proportion vector, and the functional diversity feature, the user usage characteristics of the POI clustering unit are determined.

[0066] The functional behavior category ratio vector includes multiple functional behavior category ratios. Each functional behavior category ratio represents the ratio between the number of valid user visit behavior records of that functional behavior category within the POI cluster and the total number of valid user visit behavior records within that POI cluster.

[0067] The time behavior category ratio vector includes multiple functional behavior category ratios. Each time behavior category ratio represents the ratio between the number of valid user visit behavior records of that time behavior category within the POI cluster unit and the total number of all valid user visit behavior records within that POI cluster unit.

[0068] Functional diversity is a quantitative indicator used to measure whether the functions of the business districts to be classified corresponding to the POI clustering unit are diversified or singular; the functions of the business districts to be classified include, but are not limited to, catering services, shopping services, lifestyle services, sports and leisure services, and comprehensive services, etc.

[0069] Specifically, for each POI clustering unit, when determining the user usage characteristics of the POI clustering unit, the functional behavior categories obtained from the functional exposure clustering and the time behavior categories from the time behavior clustering are combined. On the one hand, the functional behavior category proportion vector is determined based on the ratio between the number of valid user visit behavior records of each functional behavior category in the POI clustering unit and the total number of valid user visit behavior records.

[0070] On the other hand, the proportion vector of time behavior categories is determined based on the ratio between the number of valid user visit behavior records for each time behavior category within the POI clustering unit and the total number of valid user visit behavior records.

[0071] On the other hand, information entropy is used to determine the functional diversity characteristics of all valid user visit records within the POI clustering unit.

[0072] For example, if we collect 1000 valid user visit records within a month, including 400 dining-related behaviors, 300 shopping-related behaviors, 200 entertainment-related behaviors, 50 lifestyle service behaviors, and 50 comprehensive service behaviors, the percentages of each type of behavior are 0.4, 0.3, 0.2, 0.05, and 0.05, respectively. Based on these percentages, we determine the probability for calculating information entropy. , probability Substitute into the information entropy formula The information entropy was calculated to be 1.946, and this information entropy of 1.946 was taken as the functional diversity feature.

[0073] Understandably, if the proportion of each of the five functional behaviors in a business district to be classified is 0.2%, its functional diversity characteristic value will be higher (e.g., 2.322), indicating that the business district to be classified is a comprehensive business district with very balanced functions. If 90% of the visit records in a "pure shopping" business district to be classified are shopping behaviors, and the proportion of other categories is very low, then its functional diversity characteristic value will be very low (e.g., below 0.5), indicating that the business district is a specialized business district with highly concentrated functions.

[0074] Finally, based on the functional behavior category proportion vector, time behavior category proportion vector, and functional diversity characteristics of the POI cluster unit, user usage characteristics revealing the consumption structure and time rhythm of the business district corresponding to the POI cluster unit are obtained.

[0075] For example, when the functional behavior categories include dining behavior, shopping behavior, entertainment behavior, life service behavior, and comprehensive service behavior, and the time behavior clustering includes weekday noon behavior, weekday evening behavior, rest day noon behavior, rest day evening behavior, and rest day afternoon behavior, the user usage characteristics of a POI clustering unit include the functional behavior category proportion vector [0.4, 0.3, 0.2, 0.05, 0.05], the time behavior category proportion vector [0.1, 0.2, 0.2, 0.25, 0.25], and the functional diversity feature 1.946. These three are concatenated together to obtain the user usage characteristics.

[0076] The business district classification method provided by this invention determines the user usage characteristics of POI cluster units based on the proportion vectors of functional behavior categories, the proportion vectors of time behavior categories, and functional diversity characteristics. By utilizing user usage characteristics that can reveal the consumption structure and time rhythm of the business district, and preset classification characteristics, a multi-dimensional classification feature vector of POI cluster units is constructed. Finally, the multi-dimensional classification feature vector is clustered to obtain the business district classification result, thereby improving the ability of the business district classification result to depict the actual usage and service capabilities of the business district.

[0077] Based on the above embodiments, as an optional embodiment, the functional behavior categories corresponding to the multiple valid user visit records are determined based on the following method: For each valid user visit behavior record, at least two spatial buffers with different areas are defined with the visit location in the valid user visit behavior record as the geometric center of the spatial buffer. The business service POI type ratio vectors corresponding to each spatial buffer are weighted and summed according to preset weights to obtain the event-level function exposure vector of the valid user visit behavior record. Cluster the event-level function exposure vectors of each valid user visit behavior record to obtain the function behavior category corresponding to each valid user visit behavior record; The preset weight is inversely correlated with the area of ​​the spatial buffer; the commercial service POI type ratio vector corresponding to the spatial buffer is determined according to the ratio of the number of commercial service POIs of different commercial service types within the spatial buffer.

[0078] To characterize the true purpose of a user's single visit, a multi-scale spatial exposure method is used to obtain the functional behavior category corresponding to each valid user visit record.

[0079] Specifically, for each valid user visit record, at least two spatial buffers with different areas are delineated with the visit location of the valid user visit record as the geometric center of the spatial buffer. According to the requirement that the weight is inversely correlated with the area (i.e., the larger the area, the smaller the weight), the preset weight corresponding to each spatial buffer is determined.

[0080] For example, the areas of spatial buffers A, B, and C are 50m² and 50m² respectively. 2 150m 2 and 300m 2 The area ratio of the three space buffers is 1:3:6, so the preset weight ratio of space buffers A, B, and C is 6:3:1, and the preset weights of space buffers A, B, and C are 0.6, 0.3, and 0.1, respectively.

[0081] By utilizing the proportions of the number of commercial service POIs of different commercial service types (such as catering services, shopping services, lifestyle services, and sports and leisure services) within each spatial buffer, the commercial service POI type proportion vector corresponding to that spatial buffer is determined, thereby obtaining the commercial service POI type proportion vector corresponding to each spatial buffer.

[0082] For example, within a circular spatial buffer zone with a radius of 150 meters, a total of 20 commercial service POIs were found. The number of catering service POIs, shopping service POIs, lifestyle service POIs, and sports and leisure service POIs were 10, 5, 3, and 2, respectively. Then, the commercial service POI type ratio vector corresponding to this spatial buffer zone is [0.5, 0.25, 0.15, 0.1].

[0083] The proportional vectors of commercial service POI types corresponding to each spatial buffer are weighted and summed according to preset weights to obtain the event-level function exposure vector of the effective user visit behavior record.

[0084] By repeating the steps of determining the event-level function exposure vector for each valid user visit behavior record, the event-level function exposure vector for all valid user visit behavior records within the POI clustering unit can be obtained.

[0085] Further, Hellinger distance or probabilistic K-means clustering methods are used to cluster the event-level function exposure vectors of each valid user visit behavior record, resulting in function exposure vector subclusters. For each function exposure vector subcluster, based on the commercial service POI type whose cluster center probability is greater than a preset probability threshold (e.g., 50%), the dominant function category of the function exposure vector subcluster is determined. This, in turn, determines the function behavior category (dining, shopping, entertainment, lifestyle service behavior) of all valid user visit behavior records corresponding to event-level function exposure vectors within the subcluster. If no commercial service POI type has a probability greater than the preset probability threshold, it indicates that the cluster center of the function exposure vector subcluster does not have a single dominant function. In this case, the function behavior category of all valid user visit behavior records corresponding to event-level function exposure vectors within the subcluster is determined as comprehensive service behavior.

[0086] By repeating the steps of determining the functional behavior category of the valid user visit behavior records corresponding to all event-level functional exposure vectors within the functional exposure vector sub-cluster, the functional behavior category corresponding to each valid user visit behavior record within each POI clustering unit can be obtained.

[0087] Optionally, when the number of valid user visit records within a certain POI clustering unit reaches a preset scale (e.g., the number of records is greater than or equal to the preset number), the Hellinger distance between each event-level function exposure vector is not calculated. Instead, the K-means clustering method of square root probability vectors is used. That is, the event-level function exposure vectors are transformed by square root to map them into Euclidean space, and the event-level function exposure vectors of each valid user visit record are clustered using the probabilistic K-means clustering method. This can significantly reduce the computational complexity while maintaining the ability to distinguish the similarity of function probability distributions, and realize an equivalent optimization scheme in large-scale computing scenarios.

[0088] Optionally, the event-level functional exposure vectors of each valid user visit behavior record are clustered according to the second cluster number; wherein, the second cluster number is determined by drawing scree maps based on the sum of squares within each event-level functional exposure vector under different cluster numbers.

[0089] It is understood that the shape of the space buffer can be a square, a circle, a regular hexagon, etc., and this invention does not limit this.

[0090] Based on the above embodiments, as an optional embodiment, the step of delineating at least two spatial buffers of different areas with the visit location in the valid user visit behavior record as the geometric center of the spatial buffer, and weighting and summing the commercial service POI type proportion vectors corresponding to each spatial buffer according to preset weights, includes: Using the visit locations in the valid user visit records as the geometric centers of each spatial buffer, a first spatial buffer, a second spatial buffer, and a third spatial buffer are defined; the area of ​​the first spatial buffer is smaller than the area of ​​the second spatial buffer, and the area of ​​the second spatial buffer is smaller than the area of ​​the third spatial buffer. The first business service POI type ratio vector corresponding to the first spatial buffer, the second business service POI type ratio vector corresponding to the second spatial buffer, and the third business service POI type ratio vector corresponding to the third spatial buffer are weighted and summed according to the first weight, the second weight, and the third weight, respectively; the first weight is greater than the second weight, and the second weight is greater than the third weight.

[0091] For example, the areas of the first, second, and third spatial buffers are set to 50m² respectively. 2 150m 2 300m 2 The first, second, and third weights are 0.6, 0.3, and 0.1, respectively. If the first business service POI type ratio vector is [0.1, 0.2, 0.3, 0.4], the second business service POI type ratio vector is [0.4, 0.3, 0.2, 0.1], and the third business service POI type ratio vector corresponding to the third space buffer is [0.25, 0.25, 0.25, 0.25], then the weighted summation of the event-level function exposure vector is [(0.1 × 0.6 + 0.4 × 0.6 + 0.3, 0.4, ...4, 0.3, 0.4, 0.4, 0.3, 0.4, 0.4, 0. 3+0.25×0.1),(0.2×0.6+0.3×0.3+0.25×0.1),(0.3×0.6+0.2×0.3+0.25×0.1),(0.4×0.6+0.1×0.3+0.25×0.1)], which is [0.205,0.235,0.265,0.295]. The functional behavior category corresponding to this valid user visit behavior record is comprehensive service behavior.

[0092] The business district classification method provided by this invention identifies the functional behavior categories corresponding to user visit behavior records through multi-scale exposure and probabilistic spatial clustering. It can solve the problems that static POIs cannot reflect real usage preferences and that positioning errors lead to inaccurate function matching. It effectively reduces the impact of mobile phone positioning errors, making functional behavior judgment more accurate, stable, and realistic, and applicable to multi-scale spatial environments. It also has cross-regional portability.

[0093] In one embodiment, the time behavior category of multiple valid user visit records is determined based on the following method: For each valid user visit record, a time behavior vector is determined based on three time dimension indicators: date type, midpoint time of visit, and duration of stay. Standardize all valid user visit records to eliminate dimensional differences among all valid user visit records; The K-means clustering method was used to cluster the valid user visit behavior records after eliminating the differences in dimensions, and the time behavior category corresponding to each valid user visit behavior record was obtained.

[0094] Based on the above embodiments, as an optional embodiment, the preset classification features include the customer source ratio features of the POI clustering units; the customer source ratio features of the POI clustering units are determined according to the following method: The number of residential customers in the POI cluster unit is determined based on the geographical location of the business district to be classified corresponding to the POI cluster unit and the distance between the residential locations of the valid user visit records within the POI cluster unit. The number of workplace customers in the POI cluster unit is determined based on the geographical location of the business district to be classified corresponding to the POI cluster unit and the distance between the workplace locations of the valid user visit records within the POI cluster unit. The proportion of customers from residential locations and the proportion of customers from workplaces are determined based on the number of customers from residential locations and the number of customers from workplaces. The customer source ratio characteristics of the POI clustering unit are determined based on the proportion of customers from the residential location and the proportion of customers from the workplace.

[0095] Specifically, the pre-determined business district classification indicator system includes, in addition to user usage characteristics, preset classification features such as the customer source ratio of POI cluster units, and the customer source ratio is determined based on the proportion of customers from residential areas and the proportion of customers from workplaces.

[0096] On the one hand, when determining the proportion of residential customers, the residential location of the corresponding user is determined based on the valid user visit records within the POI cluster unit. The number of users whose residential location is less than a preset distance threshold (such as 3km, 5km, etc.) between the geographical location of the residential location and the geographical location of the business district to be classified corresponding to the POI cluster unit is further counted to determine the number of residential customers in the POI cluster unit.

[0097] On the other hand, when determining the proportion of customers from workplaces, the workplace location of the corresponding user is determined based on the valid user visit records within the POI cluster unit. The number of users whose workplace location is less than a preset distance threshold (such as 3km, 5km, etc.) between the geographical location of the workplace location and the geographical location of the business district to be classified corresponding to the POI cluster unit is further counted to determine the number of workplace customers in the POI cluster unit.

[0098] Furthermore, the residential user ratio of a POI cluster unit is determined by the ratio of the number of residential users to the total number of visitors to that POI cluster unit. Similarly, the workplace user ratio is determined by the ratio of the number of workplace users to the total number of visitors to that POI cluster unit. Finally, the residential and workplace user ratios are defined as the user ratio characteristics of each POI cluster unit.

[0099] It should be noted that the customer source ratio characteristics can be used to distinguish between community-based business districts, mixed-use business districts, and destination-based business districts.

[0100] The business district classification method provided by this invention introduces additional customer source ratio features on the basis of static commercial service POIs and user usage characteristic indicators. It simultaneously utilizes user usage features, customer source ratio features and preset classification features to construct multi-dimensional classification feature vectors for POI clustering units. After clustering the multi-dimensional classification feature vectors of each POI clustering unit, the business district classification result of the business district to be classified is obtained, which can further improve the ability of the business district classification result to depict the actual usage status and service capabilities of the business district.

[0101] Based on the above embodiments, as an optional embodiment, the preset classification features further include the behavioral intensity features, dwell depth features, and spatial structure features of the POI clustering units; The behavioral intensity characteristics are determined based on the scale of visits, the percentage of daily peak visits, and the average frequency of visits per person. The dwell depth feature is determined based on the average dwell time of users and the proportion of users who spend a long time. The spatial structure features are determined based on the maximum POI kernel density.

[0102] Specifically, the business district classification index system used to construct multi-dimensional classification feature vectors integrates five dimensions of index features. In addition to user usage features and customer source ratio features, it also includes three other classification features: behavioral intensity features, dwell depth features, and spatial structure features of POI clustering units.

[0103] Among them, the behavioral intensity feature reflects the consumption activity level of the business district to be classified corresponding to the POI clustering unit based on the scale of visits, the percentage of peak daily visits, and the frequency of visits per person; the dwell depth feature describes the stickiness and consumption continuity of consumers in the business district to be classified corresponding to the POI clustering unit based on the average dwell time of users and the percentage of users with long dwell time; the spatial structure feature uses the maximum value of POI kernel density to represent the commercial agglomeration level and spatial density of commercial facilities in the business district to be classified corresponding to the POI clustering unit.

[0104] When constructing the multidimensional classification feature vector corresponding to the POI clustering unit, in addition to determining user usage characteristics and customer source ratio characteristics, behavioral intensity characteristics, dwell depth characteristics, and spatial structure characteristics are also determined. These include: Firstly, based on the number of multiple valid user visit behavior records within the POI clustering unit, the visit scale representing the total number of visits attracted by the business district within the statistical period (e.g., taking a log transformation to eliminate magnitude differences) is determined; based on the proportion of valid user visit behavior records within the POI clustering unit representing visits to the business district during the busiest time of the day to the total number of valid user visit behavior records, the daily peak visit ratio is determined; based on the average number of visits per user within the business district during the statistical period, the average visit frequency per person is determined; and based on the visit scale, the daily peak visit ratio, and the average visit frequency per person, the behavioral intensity characteristics are determined.

[0105] For example, if the business district A to be classified is a business district located in a large city-level CBD, the total number of visits is determined to be 1,000,000, the peak daily visit percentage is determined to be 35% during the weekday lunch break, and the average visit frequency per person is 1.8 times / person / month, then the behavioral intensity characteristics of the business district A to be classified are [1,000,000, 35%, 1.8].

[0106] For example, if the business district B to be classified is a business district that serves as a community life center, with a total number of visits of 1,500,000, a peak daily visit rate of 15%, and an average visit frequency of 6.5 times per person per month, then the behavioral intensity characteristics of the business district B to be classified are [1,500,000, 15%, 6.5].

[0107] On the other hand, the average dwell time of users is determined based on the average duration of a single stay within the business district; the proportion of long-stay users is determined based on the proportion of valid user visit records where users stay in the business district for more than a preset duration (such as 2 hours) to the total number of valid user visit records; and the dwell depth characteristics are determined based on the average dwell time and the proportion of long-stay users.

[0108] For example, if the unclassified business district C is an experiential shopping center, the average user stay time is 160 minutes, and the percentage of users staying for a long time is 65%, then the dwell depth characteristic of the unclassified business district C is [160, 0.65].

[0109] For example, if the business district D to be classified belongs to the street-front commercial area, the average user stay time is 25 minutes, and the percentage of users staying for a long time is 5%, then the dwell depth characteristic of the business district D to be classified is [25, 0.05].

[0110] On the other hand, based on the kernel density analysis of each business service POI within the POI clustering unit, the maximum business density value within the corresponding range of businesses to be classified is obtained, that is, the maximum kernel density value of POI, which serves as a spatial structure feature.

[0111] Finally, the vectors corresponding to the user usage features, customer source ratio features, behavior intensity features, dwell depth features, and spatial structure features of the POI clustering unit are concatenated to jointly construct the multidimensional classification feature vector of the POI clustering unit.

[0112] Optionally, in application scenarios that require characterizing the dynamic changes of a business district, such as business activity monitoring, operational evaluation, and trend analysis, the preset classification features also include activity evaluation features. These features are determined based on the consumption frequency change rate, repeat visit rate, and the proportion of high-frequency users. The consumption frequency change rate is used to characterize the growth or decline trend of the frequency of visits to the business district in different periods. The repeat visit rate is used to characterize the proportion of users who visit the same business district multiple times within the statistical period, reflecting the stability of the business district's attraction to consumers. The proportion of high-frequency users is used to characterize the structural characteristics of the core consumer group of the business district.

[0113] The business district classification method provided by this invention constructs a comprehensive business district classification system that integrates user visit behavior characteristics and spatial structure characteristics, including five major categories of indicators: behavior intensity, dwell time, customer source structure, user usage characteristics, and spatial structure. This system can solve the problems of single classification dimensions, fragmented behavioral and spatial information, and lack of a unified framework. It achieves a five-dimensional integrated classification system of "scale, depth, customer source, function, and space," and improves the stability, interpretability, and cross-city reusability of the classification results.

[0114] Based on the above embodiments, as an optional embodiment, the step of spatially clustering the commercial service POIs of the target area requiring business district classification to identify multiple POI clustering units includes: Based on the first radius, the commercial service POIs are spatially clustered to obtain local high-density POI subclusters; Based on the second radius, the local high-density POI subclusters are spatially clustered to obtain POI clustering units; The second radius is larger than the first radius.

[0115] For example, the first radius is 50m and the second radius is 500m.

[0116] To improve the recognition accuracy of POI clustering units corresponding to the business districts to be classified, this invention adopts a business district recognition scheme based on dual-scale POI clustering.

[0117] Specifically, Figure 3 This is an example diagram of the locally high-density POI sub-clusters and POI clustering units provided by the present invention, as shown below. Figure 3 As shown, small-scale spatial clustering of commercial service POIs within the target area is performed based on the first radius to obtain multiple local high-density POI subclusters that reflect the building-level commercial agglomeration structure.

[0118] Further, based on the second radius, large-scale spatial clustering of each local high-density POI sub-cluster is performed to obtain multiple POI clustering units that reflect the overall space of the commercial concentration area.

[0119] It should be noted that the clustering methods used for spatial clustering of commercial service POIs and spatial clustering of local high-density POI subclusters can be DBSCAN clustering or other clustering methods; there are no restrictions on the specific methods used.

[0120] When using conventional single-scale (e.g., street block or regular grid scale) business district identification schemes to identify POI clustering units corresponding to the business district to be classified, the spatial range of the identification units is relatively large at the street block scale. This easily leads to the inclusion of some non-commercial, highly concentrated areas within the business district, resulting in the introduction of more non-target area dwell records when analyzing user visit behavior. This amplifies behavioral indicators such as visit scale and dwell time, affecting the accuracy of behavioral analysis. At the regular grid scale, the spatial range of the identification units is relatively small, easily dividing the same spatially continuous and functionally closely related business district into multiple discrete units, increasing the fragmentation of the business district and hindering the effective aggregation of user visit behavior and the hierarchical classification of business districts. Therefore, single-scale business district identification schemes often suffer from the problem of being difficult to directly serve as a stable spatial carrier for user visit event analysis.

[0121] The business district classification method provided by this invention identifies POI clustering units corresponding to business districts through two-layer spatial clustering composed of micro and macro scales. Moreover, POI clustering units can be directly used as stable spatial carriers for user visit event analysis. This solves the problems of unstable boundaries, over-segmentation, and over-merging that exist in single-scale clustering identification of business districts. It can simultaneously take into account the integrity of the business district spatial boundary and the spatial accuracy required for behavioral analysis, so that business district identification has both micro-accuracy and macro-structural consistency, thereby improving the accuracy and interpretability of business district identification.

[0122] Based on the above embodiments, as an optional embodiment, the first radius is determined according to the building radius of the target area; the second radius is determined according to the block radius of the target area.

[0123] Among them, the building radius can be determined by the average radius, median radius, and other radius data of buildings within the target area for classifying business districts as needed; the street radius can be determined by the average radius, median radius, and other radius data of streets within the target area for classifying business districts as needed.

[0124] Specifically, in the business district identification using dual-scale POI clustering, commercial complexes or individual buildings are first identified as targets. Building-scale DBSCAN clustering is performed using a smaller neighborhood radius. This involves spatially clustering commercial service POIs within the target area that needs to be classified into business districts using a first radius determined by the building radius, resulting in multiple local high-density POI subclusters to reflect the building-level commercial agglomeration structure. Then, a larger neighborhood radius is used for block-scale DBSCAN clustering. This involves spatially merging each local high-density POI subcluster using a second radius determined by the block radius to identify the overall space of the commercial concentration area, resulting in multiple POI clustering units.

[0125] The business district classification method provided by this invention identifies POI clustering units corresponding to business districts through a two-layer spatial clustering consisting of micro-building scale and macro-street scale. The first layer uses a smaller neighborhood radius to identify commercial complexes and building-level commercial clustering units. The second layer, based on the sub-clusters of the first layer, uses a larger neighborhood radius to merge spatially adjacent and highly correlated sub-clusters. This can obtain POI clustering units that reduce the noise impact of non-commercial highly concentrated areas, avoid the over-segmentation or over-merging problems that are prone to occur in single-scale clustering, thereby improving the accuracy and stability of business district identification.

[0126] Overall, this invention presents a business district classification method based on user visit behavior analysis. Through a two-layer DBSCAN clustering structure, it achieves a balance between micro-level accuracy and macro-level structural consistency, improves the stability and interpretability of business district spatial boundaries, and provides POI clustering unit business district identification results that can be directly used for user visit behavior analysis. By utilizing effective user visit behavior records within POI clustering units as core analysis units to extract user usage features, it can realistically depict users' actual usage of the business district, avoiding analytical biases caused by sparse user profiles. By determining the functional behavior categories corresponding to effective user visit behavior records through a multi-scale spatial function exposure method, it can effectively reduce the impact of positioning errors on function identification results and improve the robustness of function determination. Through a multi-dimensional classification system integrating behavior intensity, dwell depth, customer source structure, usage characteristics, and spatial structure, it achieves a systematic and reusable classification of business district types, thus possessing good cross-city applicability and transferability. It can be widely applied to cities of different sizes and types, providing unified and stable technical support for application scenarios such as urban planning, commercial site selection, commercial operation management, and consumer behavior research.

[0127] Figure 4 This is a schematic diagram of the business district type classification device provided by the present invention, as shown below. Figure 4 As shown, the business district type classification device includes, but is not limited to, a business district identification module 401, a multi-dimensional vector acquisition module 402, and a business district classification module 403.

[0128] The business district identification module 401 is used to perform spatial clustering of commercial service POIs in the target area that needs to be classified as business districts, and to identify multiple POI clustering units; wherein, one POI clustering unit corresponds to a business district to be classified.

[0129] The multidimensional vector acquisition module 402 is used to determine the user usage characteristics of each POI clustering unit based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within the statistical period; and to determine the multidimensional classification feature vector of the POI clustering unit based on the user usage characteristics and the preset classification characteristics of the POI clustering unit.

[0130] The business district classification module 403 is used to cluster the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit.

[0131] The valid user visit behavior records are determined based on user equipment signaling residency records whose visit times match the business hours of the business service POI.

[0132] It should be noted that the business district type division device provided by the present invention can execute the business district type division method described in any of the above embodiments during specific operation, which will not be elaborated in this embodiment.

[0133] The business district classification device provided by this invention introduces effective user visit behavior records that characterize actual user visits and usage behavior information on the basis of static commercial service POI data. It then uses the functional and temporal behavior categories of these records to determine the user usage characteristics of POI clustering units corresponding to the business district to be classified. Finally, it constructs multi-dimensional classification feature vectors for each POI clustering unit using these user usage characteristics and preset classification features. After clustering the multi-dimensional classification feature vectors of each POI clustering unit, the business district classification result is obtained. This enables urban business district classification analysis based on user visit behavior, effectively improving the ability of the business district classification results to depict the actual usage and service capabilities of the business district, and enhancing the explanatory power of the classification results in refined analysis and practical applications.

[0134] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 5 As shown, the electronic device may include: a processor 510, a communications interface 520, a memory 530, and a communications bus 540, wherein the processor 510, the communications interface 520, and the memory 530 communicate with each other through the communications bus 540. The processor 510 can call logical instructions in the memory 530 to execute the business district type classification method provided in any of the above embodiments. The business district type classification method includes, but is not limited to, the following steps: spatially clustering the commercial service POIs of the target area that need to be classified into business districts to identify multiple POI clustering units; wherein, one POI clustering unit corresponds to a business district to be classified; for each POI clustering unit, determining the user usage characteristics of the POI clustering unit based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within a statistical period; determining the multidimensional classification feature vector of the POI clustering unit based on the user usage characteristics and the preset classification features of the POI clustering unit; clustering the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit; wherein, the valid user visit behavior records are determined based on user equipment signaling residency records whose visit time matches the business hours of the commercial service POI.

[0135] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0136] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the business district type classification method provided in any of the above embodiments. The business district type classification method includes, but is not limited to, the following steps: spatially clustering the commercial service POIs of the target area that need to be classified into business districts, and identifying multiple POI clustering units; wherein, one POI clustering unit corresponds to one business district to be classified; for each POI clustering unit, according to the POIs within the statistical period... The functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the I cluster unit are respectively used to determine the user usage characteristics of the POI cluster unit; based on the user usage characteristics and the preset classification characteristics of the POI cluster unit, the multidimensional classification feature vector of the POI cluster unit is determined; the multidimensional classification feature vector of each POI cluster unit is clustered to obtain the business district type classification result of the business district to be classified corresponding to each POI cluster unit; wherein, the valid user visit behavior records are determined based on the user equipment signaling residency records based on the visit time conforming to the business hours of the business service POI.

[0137] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements the business district type classification method provided in any of the above embodiments. The business district type classification method includes, but is not limited to, the following steps: spatially clustering commercial service POIs in the target area that need to be classified into business districts to identify multiple POI clustering units; wherein, one POI clustering unit corresponds to one business district to be classified; for each POI clustering unit, determining the user usage characteristics of the POI clustering unit based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within a statistical period; determining the multidimensional classification feature vector of the POI clustering unit based on the user usage characteristics and the preset classification characteristics of the POI clustering unit; clustering the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit; wherein, the valid user visit behavior records are determined based on user equipment signaling residency records whose visit time matches the business hours of the commercial service POI.

[0138] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0139] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0140] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for classifying business districts, characterized in that, include: Spatial clustering is performed on commercial service points (POIs) in target areas that require business district classification to identify multiple POI clustering units; wherein, each POI clustering unit corresponds to a business district to be classified. For each POI clustering unit, the user usage characteristics of the POI clustering unit are determined based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within the statistical period; and the multidimensional classification feature vector of the POI clustering unit is determined based on the user usage characteristics and the preset classification characteristics of the POI clustering unit. Clustering is performed on the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit; The valid user visit behavior records are determined based on user equipment signaling residency records whose visit times match the business hours of the business service POI.

2. The business district type classification method according to claim 1, characterized in that, The process of determining the user usage characteristics of a POI cluster unit based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI cluster unit within the statistical period includes: The functional behavior category proportion vector of the POI clustering unit is determined based on the number of valid user visit records for each functional behavior category. Based on the number of valid user visit records for each time behavior category, determine the time behavior category proportion vector of the POI clustering unit; Based on the information entropy of multiple valid user visit records, the functional diversity characteristics of the POI clustering unit are obtained; Based on the functional behavior category proportion vector, the time behavior category proportion vector, and the functional diversity feature, the user usage characteristics of the POI clustering unit are determined.

3. The method for classifying business districts according to claim 1, characterized in that, The functional behavior categories corresponding to the multiple valid user visit records are determined based on the following method: For each valid user visit behavior record, at least two spatial buffers with different areas are defined with the visit location in the valid user visit behavior record as the geometric center of the spatial buffer. The business service POI type ratio vectors corresponding to each spatial buffer are weighted and summed according to preset weights to obtain the event-level function exposure vector of the valid user visit behavior record. Cluster the event-level function exposure vectors of each valid user visit behavior record to obtain the function behavior category corresponding to each valid user visit behavior record; The preset weight is inversely correlated with the area of ​​the spatial buffer; the commercial service POI type ratio vector corresponding to the spatial buffer is determined according to the ratio of the number of commercial service POIs of different commercial service types within the spatial buffer.

4. The method for classifying business districts according to claim 3, characterized in that, The step of defining at least two spatial buffers of different areas, using the visited locations in the valid user visit records as the geometric centers of the spatial buffers, and then weighting and summing the commercial service POI type proportion vectors corresponding to each of the spatial buffers according to preset weights, includes: Using the visit locations in the valid user visit records as the geometric centers of each spatial buffer, a first spatial buffer, a second spatial buffer, and a third spatial buffer are defined; the area of ​​the first spatial buffer is smaller than the area of ​​the second spatial buffer, and the area of ​​the second spatial buffer is smaller than the area of ​​the third spatial buffer. The first business service POI type ratio vector corresponding to the first spatial buffer, the second business service POI type ratio vector corresponding to the second spatial buffer, and the third business service POI type ratio vector corresponding to the third spatial buffer are weighted and summed according to the first weight, the second weight, and the third weight, respectively; the first weight is greater than the second weight, and the second weight is greater than the third weight.

5. The method for classifying business districts according to claim 1, characterized in that, The preset classification features include the customer source ratio features of the POI clustering units; the customer source ratio features of the POI clustering units are determined according to the following method: The number of residential customers in the POI cluster unit is determined based on the geographical location of the business district to be classified corresponding to the POI cluster unit and the distance between the residential locations of the valid user visit records within the POI cluster unit. The number of workplace customers in the POI cluster unit is determined based on the geographical location of the business district to be classified corresponding to the POI cluster unit and the distance between the workplace locations of the valid user visit records within the POI cluster unit. The proportion of customers from residential locations and the proportion of customers from workplaces are determined based on the number of customers from residential locations and the number of customers from workplaces. The customer source ratio characteristics of the POI clustering unit are determined based on the proportion of customers from the residential location and the proportion of customers from the workplace.

6. The method for classifying business districts according to claim 5, characterized in that, The preset classification features also include the behavioral intensity features, dwell depth features, and spatial structure features of the POI clustering units; The behavioral intensity characteristics are determined based on the scale of visits, the percentage of daily peak visits, and the average frequency of visits per person. The dwell depth feature is determined based on the average dwell time of users and the proportion of users who spend a long time. The spatial structure features are determined based on the maximum POI kernel density.

7. The method for classifying business districts according to claim 1, characterized in that, The process involves spatial clustering of commercial service points of interest (POIs) in the target area requiring business district classification, resulting in the identification of multiple POI clustering units, including: Based on the first radius, the commercial service POIs are spatially clustered to obtain local high-density POI subclusters; Based on the second radius, the local high-density POI subclusters are spatially clustered to obtain POI clustering units; The second radius is larger than the first radius.

8. The method for classifying business districts according to claim 7, characterized in that, The first radius is determined based on the building radius of the target area; the second radius is determined based on the block radius of the target area.

9. A business district type classification device, characterized in that, include: The business district identification module is used to spatially cluster commercial service points (POIs) in target areas that need to be classified into business districts, and to identify multiple POI clustering units; wherein, each POI clustering unit corresponds to a business district to be classified. The multidimensional vector acquisition module is used to determine the user usage characteristics of each POI clustering unit based on the functional behavior category and time behavior category corresponding to multiple valid user visit behavior records of the POI clustering unit within the statistical period; and to determine the multidimensional classification feature vector of the POI clustering unit based on the user usage characteristics and the preset classification characteristics of the POI clustering unit. The business district classification module is used to cluster the multidimensional classification feature vectors of each POI clustering unit to obtain the business district type classification result of the business district to be classified corresponding to each POI clustering unit; The valid user visit behavior records are determined based on user equipment signaling residency records whose visit times match the business hours of the business service POI.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the business district type division method as described in any one of claims 1 to 8.

Images