Position Selection Method, Device, Computer Equipment and Computer Readable Storage Medium

Abstract

The present application discloses a location selection method, device, computer device, and computer-readable storage medium, which relates to the field of Internet technologies. The multi-dimensional data features of candidate location points are fused and scored by using a location point feature model, and the impact of location selection on the overall distribution scheduling is comprehensively considered, avoiding single-index decision-making. Not only is the location selection efficiency high, but also the globality, accuracy, and effectiveness of location selection can be ensured. The method includes: determining a plurality of candidate location points, and obtaining the location point features of each candidate location point among the plurality of candidate location points; scoring the location point features of each candidate location point based on the location point feature model to obtain a plurality of location scores of the plurality of candidate location points; sorting the plurality of candidate location points according to the plurality of location scores to obtain a sorting result, and selecting a target location point in the sorting result for pushing.

Description

Technical Field

[0001] This application relates to the field of Internet technologies, and particularly to a method and device for location selection, a computer device, and a computer-readable storage medium. Background Art

[0002] With the continuous development of Internet technologies, online shopping has become an essential part of people's lives and work. People can purchase goods such as clothing online and also order takeaways. The goods ordered online usually need to be delivered to customers. As the last-mile delivery equipment directly facing customers, its forms can be diverse, such as stations, third-party collection points, express cabinets, takeaway cabinets, etc. Therefore, how to select the installation location of the last-mile delivery equipment has become an important issue.

[0003] In related technologies, when performing location selection, the staff responsible for site selection go to the actual scenarios to collect information such as site signals, scenario areas, equipment placement conditions, etc., investigate and collect parameters such as the number of permanent residents, the number of buildings, and the number of floors in the scenario, and manually determine at which location in the scenario to place the equipment based on the collected information and parameters.

[0004] In the process of implementing this application, the applicant found that the related technologies have at least the following problems:

[0005] Manually performing location selection requires a large amount of manpower, is time-consuming and laborious. Not only is the efficiency of location selection low, but it is also difficult to ensure the globality, accuracy, and effectiveness of location selection. Summary of the Invention

[0006] In view of this, this application provides a method and device for location selection, a computer device, and a computer-readable storage medium, mainly aiming to solve the problem that currently, manually performing location selection requires a large amount of manpower, is time-consuming and laborious, not only is the efficiency of location selection low, but it is also difficult to ensure the globality, accuracy, and effectiveness of location selection.

[0007] According to a first aspect of this application, there is provided a method for location selection, the method including:

[0008] Determine a plurality of candidate location points, and obtain the location point features of each candidate location point among the plurality of candidate location points, where the location point features include an index for indicating the environmental state of the candidate location point and an index for indicating the estimated delivery state after placing equipment at the candidate location point;

[0009] Score the location point features of each candidate location point based on the location point feature model to obtain multiple location point scores for the multiple candidate location points. The location point feature model is a model trained with the location points of installed devices as samples for scoring location point features.

[0010] Sort the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and select a target location point from the sorting result for pushing.

[0011] Optionally, before determining the multiple candidate location points and obtaining the location point features of each candidate location point among the multiple candidate location points, the method further includes:

[0012] Determine multiple installed devices, use the location points where the multiple installed devices are located as multiple sample location points, and use the location point features of the multiple sample location points as multiple sample features.

[0013] Split each sample feature in the multiple sample features into multiple feature items according to the described feature content, query the feature item scores corresponding to each split feature item, and label each feature item with the feature item scores.

[0014] Group the feature items with the same described feature content into the same group to obtain multiple feature groups.

[0015] Train the multiple feature groups using the logistic regression algorithm to obtain the location point feature model.

[0016] Optionally, determining the multiple candidate location points and obtaining the location point features of each candidate location point among the multiple candidate location points includes:

[0017] Determine the device to be installed and the area to be installed, and obtain the floor area of the device to be installed.

[0018] Query the area map of the area to be installed, and determine multiple candidate areas on the area map whose areas meet the floor area of the device as the multiple candidate location points.

[0019] For each candidate location point among the multiple candidate location points, count the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate location point as multiple feature items, and combine the multiple feature items as the location point feature of the candidate location point.

[0020] Optionally, the counting the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate location point as multiple feature items includes:

[0021] Taking the candidate location point as the center, circle a statistical area with a preset area, count the number of historical orders in the statistical area within a specified historical time period, and query the delivery coverage range of the candidate location point, and use the number of historical orders and the delivery coverage range as the spatio-temporal feature data;

[0022] Generate a delivery difficulty prediction index for the statistical area according to the geographical location parameters of the statistical area, and determine a time prediction index for the candidate location point according to the delivery times of multiple historical orders that occurred in the statistical area within the specified historical time period, and use the delivery difficulty prediction index and the time prediction index as the delivery feature data;

[0023] Query the resource types of each historical order in the multiple historical orders, calculate the occurrence ratio of each resource type in the multiple historical orders, and obtain the device recognition rate of the device to be placed, and use the occurrence ratio of each resource type and the device recognition rate as the preference feature data;

[0024] Obtain the device cost data, installation cost data, and maintenance cost data of the device to be placed as the cost feature data;

[0025] Query the historical placement information of the statistical area, count the abnormal frequency of abnormalities in the historical placement information, and use the abnormal frequency as the abnormal feature data;

[0026] Respectively use the spatio-temporal feature data, the delivery feature data, the preference feature data, the cost feature data, and the abnormal feature data as a feature item to obtain the multiple feature items.

[0027] Optionally, the generating a delivery difficulty prediction index for the statistical area according to the geographical location parameters of the statistical area includes:

[0028] Query the geographical parameters of the statistical area, determine the road network information covering the candidate location point in the geographical parameters, and determine the road network level corresponding to the road network information;

[0029] Determine multiple regional buildings in the statistical area, obtain the building parameters of each regional building in the multiple regional buildings, and determine the building level corresponding to the building parameters of each regional building;

[0030] Perform a weight calculation on the road network level and the building level, and use the calculated value as the delivery difficulty prediction index.

[0031] Optionally, determining the time estimation index of the candidate location points according to the delivery times of multiple historical orders that occurred in the specified historical time period in the area to be counted includes:

[0032] Perform the following processing on each of the multiple historical orders: query the merchant location information, historical delivery time, and delivery distance of the historical order, calculate the ratio between the delivery distance and the historical delivery time as the delivery speed of the historical order, count the location distance between the merchant location information and the candidate location point, calculate the ratio between the location distance and the delivery speed as the estimated delivery time, and calculate the time difference between the historical delivery time and the estimated delivery time;

[0033] Obtain the multiple time differences of the multiple historical orders, perform logistic regression calculation on the multiple time differences, and use the calculated numerical result as the time estimation index.

[0034] Optionally, scoring the location point features of each candidate location point based on the location point feature model to obtain multiple location point scores of the multiple candidate location points includes:

[0035] For each candidate location point, split the location point features of the candidate location point into multiple features to be scored;

[0036] Input the multiple features to be scored into the location point feature model respectively. The location point feature model scores the multiple features to be scored and performs logistic regression processing on the obtained multiple scores to obtain and output the location point score of the candidate location point;

[0037] Obtain the location point scores of each candidate location point output by the location point feature model to obtain the multiple location point scores.

[0038] Optionally, selecting a target location point for pushing in the sorting result includes:

[0039] When the sorting result indicates that the multiple location point scores are sorted from high to low, select at least one first location point score at the head of the sorting result, determine the target location point corresponding to each first location point score among the multiple candidate location points, and use the at least one first location point score to label the corresponding target location point, and push the labeled target location point;

[0040] When the sorting result indicates that the multiple position scores are sorted from low to high, at least one second position score at the end of the sorting result queue is selected, the target position point corresponding to each second position score among the multiple candidate position points is determined, and the corresponding target position point is labeled using the at least one second position score, and the labeled target position point is pushed.

[0041] According to a second aspect of the present application, a position selection device is provided, and the device includes:

[0042] An acquisition module, configured to determine multiple candidate position points, and acquire the position point features of each candidate position point among the multiple candidate position points, where the position point features include an index for indicating the environmental state of the candidate position point and an index for indicating the estimated delivery state after placing a device at the candidate position point;

[0043] A scoring module, configured to score the position point features of each candidate position point based on a position point feature model, to obtain multiple position scores of the multiple candidate position points, where the position point feature model is a model trained with the position points of the installed devices as samples for scoring position point features;

[0044] A selection module, configured to sort the multiple candidate position points according to the multiple position scores to obtain a sorting result, and select a target position point in the sorting result for pushing.

[0045] Optionally, the device further includes:

[0046] A determination module, configured to determine multiple installed devices, use the position points where the multiple installed devices are located as multiple sample position points, and use the position point features of the multiple sample position points as multiple sample features;

[0047] A labeling module, configured to split each sample feature among the multiple sample features into multiple feature items according to the described feature content, query the feature item scores corresponding to each split feature item, and label each feature item using the feature item scores;

[0048] A grouping module, configured to divide the feature items with the same described feature content into the same group to obtain multiple feature groups;

[0049] A training module, configured to train the multiple feature groups using a logistic regression algorithm to obtain the position point feature model.

[0050] Optionally, the obtaining module is configured to determine the device to be placed and the area to be placed, obtain the floor area of the device to be placed; query the area map of the area to be placed, and determine multiple candidate areas on the area map whose areas meet the floor area of the device as the multiple candidate location points; for each candidate location point among the multiple candidate location points, count the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate location point as multiple feature items, and combine the multiple feature items as the location point feature of the candidate location point.

[0051] Optionally, the obtaining module is configured to circle a statistical area with a preset area centered on the candidate location point, count the historical order quantity in the statistical area within a specified historical time period, and query the delivery coverage range of the candidate location point, and use the historical order quantity and the delivery coverage range as the spatio-temporal feature data; generate a delivery difficulty estimation index for the statistical area according to the geographical location parameters of the statistical area, and determine the time estimation index of the candidate location point according to the delivery time of multiple historical orders occurring in the statistical area within the specified historical time period, and use the delivery difficulty estimation index and the time estimation index as the delivery feature data; query the resource type of each historical order among the multiple historical orders, calculate the appearance ratio of each resource type in the multiple historical orders, and obtain the device recognition degree of the device to be placed, and use the appearance ratio of each resource type and the device recognition degree as the preference feature data; obtain the device cost data, installation cost data, and maintenance cost data of the device to be placed as the cost feature data; query the historical placement information of the statistical area, count the abnormal frequency of abnormalities in the historical placement information, and use the abnormal frequency as the abnormal feature data; respectively use the spatio-temporal feature data, the delivery feature data, the preference feature data, the cost feature data, and the abnormal feature data as a feature item to obtain the multiple feature items.

[0052] Optionally, the obtaining module is configured to query the geographical parameters of the statistical area, determine the road network information covering the candidate location point in the geographical parameters, and determine the road network level corresponding to the road network information; determine multiple regional buildings in the statistical area, obtain the building parameters of each regional building among the multiple regional buildings, and determine the building level corresponding to the building parameters of each regional building; perform a weight calculation on the road network level and the building level, and use the calculated value as the delivery difficulty estimation index.

[0053] Optionally, the obtaining module is configured to perform the following processing on each of the multiple historical orders: query the merchant location information, historical delivery time, and delivery distance of the historical order, calculate the ratio between the delivery distance and the historical delivery time as the delivery speed of the historical order, count the location distance between the merchant location information and the candidate location points, calculate the ratio between the location distance and the delivery speed as the estimated delivery time, and calculate the time difference between the historical delivery time and the estimated delivery time; obtain the multiple time differences of the multiple historical orders, perform logistic regression calculation on the multiple time differences, and use the calculated numerical result as the time estimation index.

[0054] Optionally, the scoring module is configured to, for each candidate location point, split the location point features of the candidate location point into multiple features to be scored; input the multiple features to be scored into the location point feature model respectively, and the location point feature model scores the multiple features to be scored, and performs logistic regression processing on the obtained multiple scores to obtain and output the location point score of the candidate location point; obtain the location point scores of each candidate location point output by the location point feature model to obtain the multiple location point scores.

[0055] Optionally, the selection module is configured to, when the sorting result indicates that the multiple location point scores are sorted from high to low, select at least one first location point score ranked at the head of the sorting result, determine the target location point corresponding to each first location point score among the multiple candidate location points, and label the corresponding target location point with the at least one first location point score, and push the labeled target location point; when the sorting result indicates that the multiple location point scores are sorted from low to high, select at least one second location point score ranked at the end of the sorting result, determine the target location point corresponding to each second location point score among the multiple candidate location points, and label the corresponding target location point with the at least one second location point score, and push the labeled target location point.

[0056] According to a third aspect of the present application, there is provided a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the method according to any one of the first aspects are implemented.

[0057] According to a fourth aspect of the present application, there is provided a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method according to any one of the first aspects are implemented.

[0058] With the above technical solution, a location selection method, device, computer device, and computer-readable storage medium provided by the present application determine multiple candidate location points, obtain the location point features of each candidate location point, including indicators for indicating the environmental state of the candidate location point and indicators for indicating the estimated delivery status after placing a device at the candidate location point, score the location point features of each candidate location point based on a location point feature model to obtain multiple location scores, sort the multiple candidate location points according to the multiple location scores to obtain a sorting result, and select a target location point for pushing in the sorting result. By using the location point feature model to fuse and score the multi-dimensional data features of the candidate location points, and comprehensively considering the impact of location selection on the overall delivery scheduling, single-index decision-making is avoided. Not only is the location selection efficiency high, but also the globality, accuracy, and effectiveness of location selection can be guaranteed.

[0059] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of the present application more obvious and understandable, the specific embodiments of the present application are specifically exemplified below. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present application. And throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0061] Figure 1 shows a schematic flowchart of a location selection method provided by an embodiment of the present application;

[0062] Figure 2A shows a schematic flowchart of a location selection method provided by an embodiment of the present application;

[0063] Figure 2B shows a schematic architecture diagram of a location selection decision-making system provided by an embodiment of the present application;

[0064] Figure 2C shows a schematic flowchart of a location selection method provided by an embodiment of the present application;

[0065] Figure 2D shows a schematic flowchart of a location selection method provided by an embodiment of the present application;

[0066] Figure 3 shows a schematic structural diagram of a location selection device provided by an embodiment of the present application;

[0067] Figure 4The figure shows a schematic structural diagram of a computer device provided by an embodiment of the present application. Detailed implementation manners

[0068] Hereinafter, exemplary embodiments of the present application will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present application can be more thoroughly understood and the scope of the present application can be fully conveyed to those skilled in the art.

[0069] An embodiment of the present application provides a position selection method, as Figure 1 shown, the method includes:

[0070] 101. Determine a plurality of candidate position points, and obtain the position point features of each candidate position point among the plurality of candidate position points.

[0071] The applicant recognizes that the end-delivery device plays an important role as a delivery service directly facing customers, especially in scenarios such as closed campuses, schools, and scenarios that advocate contactless pick-up. How to select a reasonable device placement point to maximize the utilization of the placed devices as much as possible has become an urgent problem to be solved. Therefore, the present application proposes a position selection method to train a position point feature model, which makes decision scores based on the fusion of multi-dimensional data features to achieve efficient and reasonable site selection and installation of devices.

[0072] First, in the embodiment of the present application, it is necessary to pre-select a plurality of position points as a plurality of candidate position points. Taking the device to be installed as an intelligent meal pick-up cabinet as an example, it is possible to determine a building where the intelligent meal pick-up cabinet is not installed, and take a certain position on a certain floor of the building as a candidate position point, or determine a school where the intelligent meal pick-up cabinet is not installed, and take a certain position near the school gate as a candidate position point, etc. It should be noted that the selection of candidate position points can be automatically executed by the server, that is, the server identifies the electronic map of a certain jurisdiction or a certain school, determines a suitable position on the electronic map and outputs it as a candidate position point, or the candidate position points can also be manually determined by the staff. The staff evaluates the area, selects a position point considered suitable as a candidate position point and conducts subsequent further evaluations. The present application does not specifically limit the manner of determining a plurality of candidate position points. In the actual application process, closed scenarios such as schools and communities can be used as the main body, and a plurality of positions located beside different buildings or different roads in the main body can be selected as candidate position points, so as to determine the most suitable position to place the device in schools and communities through the technical solution in the embodiment of the present application.

[0073] After determining multiple candidate location points, in order to evaluate the multi-dimensional data features of the candidate location points, it is necessary to obtain the location point features of each candidate location point among the multiple candidate location points, which is convenient for subsequent scoring of the candidate location points based on the location point features to determine whether the candidate location points are suitable as the location points for placing the device. Among them, the location point features include feature data with multiple dimensions, specifically including indicators for indicating the environmental state where the candidate location point is located and indicators for indicating the estimated delivery status after placing the device at the candidate location point, so as to take into account both the impact of site selection on the overall delivery scheduling and the needs of users in the site selection process.

[0074] 102. Score the location point features of each candidate location point based on the location point feature model to obtain multiple location point scores for the multiple candidate location points.

[0075] In the embodiment of the present application, after obtaining the location point features of each candidate location point among the multiple candidate location points, since the location point feature model is trained in the embodiment of the present application, therefore, the location point features of each candidate location point can be scored based on the location point feature model to obtain multiple location point scores for the multiple candidate location points. Among them, the location point feature model is a model trained with the location points of the installed devices as samples for scoring the location point features. In this way, the location point feature model can comprehensively consider the multi-dimensional features of each candidate location point and obtain the location point scores of each candidate location point output by the location point feature model. Referring to the high and low of the location point scores, it is possible to more intuitively determine which candidate location point is more suitable as the installation address of the device and achieve efficient site selection of the device.

[0076] 103. Sort the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and select a target location point in the sorting result for pushing.

[0077] Since the higher the score, the more suitable it is as the installation address of the device, therefore, it is necessary to sort the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and select a target location point in the sorting result for pushing. Specifically, the location point scores can be sorted from high to low or from low to high. After sorting, select the location points with higher scores for pushing for the reference of the staff.

[0078] It should be noted that the number of target location points to be pushed can actually be set manually to one or more. If it is one, then push the target location point with the highest score; if it is more than one, then determine the specific recommended number, select the recommended number of location points with higher scores as the target location points for pushing, and in order to enable the staff to know the high and low scores of these location points during pushing, the location point scores corresponding to the location points can be used for marking to clearly indicate the advantages and disadvantages of these location points.

[0079] The method provided by the embodiment of the present application determines multiple candidate location points, obtains the location point features of each candidate location point, including indicators for indicating the environmental state of the candidate location point and indicators for indicating the estimated delivery state after placing the device at the candidate location point, scores the location point features of each candidate location point based on the location point feature model to obtain multiple location point scores, sorts the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and selects a target location point for pushing in the sorting result. By using the location point feature model to fuse and score the multi-dimensional data features of the candidate location points, and comprehensively considering the impact of site selection on the overall delivery scheduling, single-index decision-making is avoided. Not only is the site selection efficiency high, but also the globality, accuracy, and effectiveness of the site selection can be guaranteed.

[0080] Further, as a refinement and extension of the specific implementation manner of the above embodiment, in order to completely describe the specific implementation process of this embodiment, the embodiment of the present application provides another location selection method, as Figure 2A shown, this method includes:

[0081] 201. Train the location point feature model.

[0082] Since the embodiment of the present application realizes multi-dimensional feature fusion by means of model training, it is necessary to implement the training of the location point feature model for fusing multi-dimensional features. The technical solution described in the embodiment of the present application can be executed by a server. A site selection decision-making system can be set in the server, as Figure 2B shown. The site selection decision-making system can include two modules, namely a model training module and an online decision-making module. Among them, the model training module is used to collect sample data, perform feature extraction and training on the sample data to obtain a location point feature model, and connect to the online decision-making module to provide the location point feature model to the online decision-making module. The online decision-making module stores the location point feature model in the DB (Data Base), so that the online decision-making module can use this model to score features; in addition, the model training module will continuously collect new sample data and continuously update the location point feature model to ensure the accuracy of the location point feature model scoring. The online decision-making module is used to score the determined multiple candidate location points. After the server determines or receives the candidate location points selected by the staff, the location point features of the candidate location points can be input into the online decision-making module. The online decision-making module uses the location point feature model to score each candidate location point, and then determines the device placement location point according to the score; in addition, the online decision-making module can output the location point score and the determined device placement location point based on the interface with the front end for the staff to refer to.

[0083] When training the location point feature model, it is necessary to select some devices that have been successfully installed and started operating as samples for training the model. The specific process of training the location point feature model is as follows:

[0084] First, determine multiple installed devices, use the location points where the multiple installed devices are located as multiple sample location points, and use the location point features of the multiple sample location points as multiple sample features. Subsequently, split each sample feature in the multiple sample features into multiple feature items according to the described feature content, query the feature item scores corresponding to each split feature item, and label each feature item using the feature item scores. Among them, the feature item scores can be set in advance by the staff according to the operating status of the installed devices. Then, divide the feature items with the same described feature content into the same group to obtain multiple feature groups, and use the logistic regression algorithm to train the multiple feature groups to obtain the location point feature model. In this way, a location point feature model constructed using the Logistic Regression (LR) model can be obtained for subsequent feature fusion.

[0085] It should be noted that since the subsequent scoring of candidate location points is based on the location point features, and the location point features include indicators for indicating the environmental status of the candidate location points and indicators for indicating the estimated delivery status after installing devices at the candidate location points, when collecting the location point features of the sample location points as sample features, it is also possible to collect the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the sample location points described in step 203 below as multiple feature items, and use these collected feature items as the sample features provided by the sample location point for model training. The specific collection process is as described in step 203 and will not be elaborated here.

[0086] 202. Determine the device to be installed and the area to be installed, obtain the floor area of the device to be installed, query the area map of the area to be installed, and determine multiple candidate areas on the area map whose areas meet the floor area of the device as multiple candidate locations.

[0087] In the embodiments of the present application, since it is necessary to select a suitable location to place the device in the area where the device is not placed, and the floor areas of different devices are different, it is necessary to first determine the device to be placed and the area to be placed, obtain the floor area of the device to be placed, query the area map of the area to be placed, and determine multiple candidate areas on the area map whose areas meet the floor area of the device as multiple candidate positions. For example, assume that the device to be installed is a meal pickup cabinet and the area to be placed is School A. Then, select multiple positions on the map of School A that can accommodate the meal pickup cabinet as candidate positions. It should be noted that the candidate positions can be selected manually or by the server automatically identifying the electronic map of the area. The present application does not make specific limitations on this.

[0088] 203. For each candidate position point among the multiple candidate position points, count the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate position point as multiple feature items, and combine the multiple feature items as the position point feature of the candidate position point.

[0089] In the embodiments of the present application, after determining multiple candidate position points, start collecting the data of each candidate position point as the position point feature, so as to score the candidate position points by fusing the position point features subsequently. Among them, the position point feature is a multi-dimensional feature including an index for indicating the environmental state of the candidate position point and an index for indicating the estimated delivery state after placing the device at the candidate position point; specifically, the position point feature may include feature items such as spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data, that is, combine these feature items as the position point feature of the candidate position point. The following describes the acquisition methods of these feature items:

[0090] The spatio-temporal feature data can include two parts. One part is related to time, mainly data such as the historical order quantity within a certain range around the candidate position point, and the other part is related to space, mainly the spatial coverage index of the candidate position point. Specifically, when collecting the spatio-temporal feature data, it is necessary to circle a preset area to be counted with the candidate position point as the center, count the historical order quantity in the area to be counted within the specified historical time period, and query the delivery coverage range of the candidate position point, and use the historical order quantity and the delivery coverage range as the spatio-temporal feature data. For example, for candidate position point A, circle a range of 5 kilometers with A as the center as the area to be counted, and count that the historical order quantity in this area in the past 30 days is 3000, and the delivery coverage range of A can reach 6 kilometers. Then, use 3000 orders and 6 kilometers as the spatio-temporal feature data of A. Among them, the above example takes the specified historical time period as the past 30 days for illustration. In the actual application process, the specified historical time period can be the past 7 days, the past 15 days, etc. The present application does not make specific limitations on this.

[0091] The delivery feature data mainly includes the estimated delivery difficulty index of candidate location points and the estimated time index of candidate location points for delivery scheduling. When collecting the delivery feature data, on the one hand, it is necessary to generate the estimated delivery difficulty index according to the geographical location parameters of the area to be counted. The estimated delivery difficulty index is actually mainly used to consider the surrounding environment of the candidate location points. The specific process of generating the estimated delivery difficulty index is as follows: First, query the geographical parameters of the area to be counted, determine the road network information covering the candidate location points in the geographical parameters, and determine the road network level corresponding to the road network information. Information such as the number of road networks that can reach the candidate location points and whether there is road surface construction can be used as the road network information. Then, it is necessary to determine multiple regional buildings in the area to be counted, obtain the building parameters of each regional building among the multiple regional buildings, and determine the building level corresponding to the building parameters of each regional building. Information such as building height, number of building floors, and whether there is an elevator can be used as the building parameters. Finally, calculate the weights of the road network level and the building level, and use the calculated value as the estimated delivery difficulty index. It should be noted that in the actual application process, statistical algorithms such as logistic regression algorithms can be used for the fusion calculation of the road network level and the building level, or information such as the number of road networks that can reach the candidate location points, whether there is road surface construction, building height, number of building floors, and whether there is an elevator determined can also be directly used as the estimated delivery difficulty index for subsequent feature fusion, without fusing the above information into a specific value.

[0092] On the other hand, it is necessary to determine the time estimation index of the candidate location point based on the delivery times of multiple historical orders that occurred in the specified historical time period in the area to be counted, and then use the delivery difficulty estimation index and the time estimation index as delivery feature data. That is to say, the time estimation index can be obtained from the analysis of the delivery times of historical orders. The time estimation index is mainly used to predict the improvement of the delivery time for the entire area to be counted after installing the device at the candidate location point. The specific process of generating the time estimation index is as follows: For each of the multiple historical orders, first, query the merchant location information, historical delivery time, and delivery distance of the historical order, and calculate the ratio between the delivery distance and the historical delivery time as the delivery speed of the historical order. Then, count the location distance between the merchant location information and the candidate location point, calculate the ratio between the location distance and the delivery speed as the estimated delivery time, and calculate the time difference between the historical delivery time and the estimated delivery time. For example, if the historical delivery time of historical order A is 20 minutes and the delivery distance is 2 kilometers, and the calculated delivery speed is 0.1 kilometer per minute, then when the distance between the merchant location information of this historical order and the candidate location point is counted as 1.8 kilometers, the estimated delivery time can be calculated as 18 minutes, and the time difference corresponding to this historical order is 2 minutes. In fact, 2 minutes can be regarded as the time that can be saved on this historical order if the candidate location point is used as the final delivery address. In this way, by calculating each historical order separately, multiple time differences of multiple historical orders can be obtained, and a logical regression calculation is performed on the multiple time differences, and the calculated numerical result is used as the time estimation index of the candidate location point.

[0093] The preference feature data mainly includes the statistical index of the resource types of the resources purchased by users within a certain range near the candidate location point and the index of the user recognition of the device to be installed. Specifically, when collecting the preference feature data, it is necessary to query the resource types ordered in each of the multiple historical orders and calculate the occurrence ratio of each resource type in the multiple historical orders. For example, for the pick-up cabinet used to hold takeout, it is possible to count what types of meals are preferred by nearby users in the past 30 days and the proportions of various types of meals. Then, obtain the user recognition of the device to be installed, and use the occurrence ratio of each resource type and the user recognition of the device as the preference feature data.

[0094] The cost feature data mainly includes the cost of the device to be installed itself, the cost required to install the device to be installed at the candidate location point, and the cost required to maintain the device to be installed at the candidate location point. Therefore, it is sufficient to obtain the device cost data, installation cost data, and maintenance cost data of the device to be installed as the cost feature data.

[0095] The abnormal feature data is mainly used to indicate the potential risks that may exist when installing a device at a candidate location point, such as whether there is a risk of prohibited installation at the candidate location point, the probability of being damaged, etc. Specifically, the historical installation information of the area to be counted can be queried, and the abnormal frequency of the abnormalities that occurred in the historical installation information can be counted. The abnormal frequency is used as the abnormal feature data. For example, query the number of device complaints and device maintenance received in the area to be counted in the past 30 days as the abnormal frequency, so as to determine whether the area to be counted is suitable for installing the device.

[0096] In this way, after determining the spatio-temporal feature data, distribution feature data, preference feature data, cost feature data, and abnormal feature data through the above process, these data can be used as a feature item respectively to obtain multiple feature items, and the multiple feature items are combined as the location point features of the candidate location point to achieve multi-dimensional feature collection of the candidate location point.

[0097] 204. Input the location point features of each candidate location point into the location point feature model, obtain the location point scores of each candidate location point output by the location point feature model, and get multiple location point scores.

[0098] In the embodiment of the present application, after collecting the location point features of each candidate location point, the location point features of each candidate location point can be input into the location point feature model to obtain the location point scores of each candidate location point output by the location point feature model, and get multiple location point scores.

[0099] Among them, since the location point features actually include multiple feature items such as spatio-temporal feature data, distribution feature data, preference feature data, cost feature data, and abnormal feature data, for each candidate location point, the location point features of the candidate location point can be first split into multiple feature items to be scored, and the multiple feature items to be scored are respectively input into the location point feature model. The location point feature model scores the multiple feature items to be scored, and performs logistic regression processing on the obtained multiple scores to obtain the location point score of the candidate location point and output it. Specifically, see Figure 2C , the spatio-temporal feature data, distribution feature data, preference feature data, cost feature data, and abnormal feature data are respectively input into the location point feature model as a feature item, so that the location point feature model scores these feature items respectively, and performs logistic regression calculation on the obtained scores to obtain a final score value as the location point score of the candidate location point.

[0100] 205. Sort the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and select a target location point in the sorting result for pushing.

[0101] In the embodiments of the present application, after obtaining multiple location scores, sort the multiple candidate location points according to the multiple location scores to obtain a sorting result, and select a target location point from the sorting result for pushing. Among them, when the sorting result indicates that the multiple location scores are sorted from high to low, select at least one first location score ranked at the head of the sorting result, determine the target location point corresponding to each first location score among the multiple candidate location points, and use the at least one first location score to label the corresponding target location point, and push the labeled target location point; when the sorting result indicates that the multiple location scores are sorted from low to high, select at least one second location score ranked at the end of the sorting result, determine the target location point corresponding to each second location score among the multiple candidate location points, and use the at least one second location score to label the corresponding target location point, and push the labeled target location point. That is, ensure that the candidate location point with the highest score is pushed as the target location point.

[0102] In addition, in the actual application process, the candidate location point with the highest score can also be directly output as the location point for installing the device, so that the staff can directly arrange the installation of the device, improving the device installation efficiency. The present application does not limit the specific method of selecting and pushing the location point.

[0103] To sum up, the process of the location selection method described in the embodiments of the present application is summarized as follows:

[0104] As Figure 2D shown, recall multiple candidate location points, obtain the multi-dimensional location point features of each candidate location point, fuse the multi-dimensional location point features of the candidate location points through a location point feature model, calculate the location scores of each candidate location point respectively, and sort the multiple candidate location points according to the score, so as to realize the pushing of the location points. In this way, an appropriate location point can be selected through the online service for placing the device, and the online service can be used to provide support for the site selection decision, which not only reduces the workload of the previous manual site selection, but also comprehensively considers multi-dimensional features such as spatio-temporal features, distribution features, preference features, cost features, and abnormal features, avoiding single-index decision-making and making the decision better.

[0105] The method provided by the embodiments of the present application determines multiple candidate location points, obtains the location point features of each candidate location point, including indicators for indicating the environmental state of the candidate location point and indicators for indicating the estimated delivery status after placing the device at the candidate location point, scores the location point features of each candidate location point based on the location point feature model, obtains multiple location point scores, sorts the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and selects a target location point for pushing in the sorting result. By using the location point feature model to fuse and score the multi-dimensional data features of the candidate location points and comprehensively considering the impact of site selection on the overall delivery scheduling, single-index decision-making is avoided. It not only has high site selection efficiency but also can ensure the globality, accuracy, and effectiveness of site selection.

[0106] Further, as Figure 1 a specific implementation of the method, the embodiments of the present application provide a location selection device, as Figure 3 shown. The device includes: an acquisition module 301, a scoring module 302, and a selection module 303.

[0107] The acquisition module 301 is used to determine multiple candidate location points and obtain the location point features of each candidate location point among the multiple candidate location points. The location point features include indicators for indicating the environmental state of the candidate location point and indicators for indicating the estimated delivery status after placing the device at the candidate location point;

[0108] The scoring module 302 is used to score the location point features of each candidate location point based on the location point feature model to obtain multiple location point scores of the multiple candidate location points. The location point feature model is a model trained with the location points of the installed devices as samples for scoring the location point features;

[0109] The selection module 303 is used to sort the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and select a target location point for pushing in the sorting result.

[0110] In a specific application scenario, the device further includes:

[0111] a determination module, used to determine multiple installed devices, use the location points where the multiple installed devices are located as multiple sample location points, and use the location point features of the multiple sample location points as multiple sample features;

[0112] a marking module, used to split each sample feature among the multiple sample features into multiple feature items according to the described feature content, query the feature item scores corresponding to each split feature item, and mark each feature item with the feature item scores;

[0113] A grouping module, configured to divide feature items with consistent described feature content into the same group, obtaining multiple feature groups;

[0114] A training module, configured to train the multiple feature groups by using a logistic regression algorithm to obtain the location point feature model.

[0115] In a specific application scenario, the obtaining module 301 is configured to determine a device to be placed and a placement area, obtain the floor area of the device to be placed; query the area map of the placement area, and determine multiple candidate areas on the area map whose areas meet the floor area of the device as the multiple candidate location points; for each candidate location point among the multiple candidate location points, count the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate location point as multiple feature items, and combine the multiple feature items as the location point feature of the candidate location point.

[0116] In a specific application scenario, the obtaining module 301 is configured to circle a to-be-counted area with a preset area centered on the candidate location point, count the number of historical orders in the to-be-counted area during a specified historical time period, and query the delivery coverage range of the candidate location point, and use the number of historical orders and the delivery coverage range as the spatio-temporal feature data; generate a delivery difficulty estimation index for the to-be-counted area according to the geographical location parameters of the to-be-counted area, and determine a time estimation index for the candidate location point according to the delivery times of multiple historical orders that occurred in the to-be-counted area during the specified historical time period, and use the delivery difficulty estimation index and the time estimation index as the delivery feature data; query the resource types ordered in each of the multiple historical orders, calculate the occurrence ratio of each resource type in the multiple historical orders, and obtain the device recognition degree of the device to be placed, and use the occurrence ratio of each resource type and the device recognition degree as the preference feature data; obtain the device cost data, installation cost data, and maintenance cost data of the device to be placed as the cost feature data; query the historical placement information of the to-be-counted area, count the abnormal frequency of abnormalities that occurred in the historical placement information, and use the abnormal frequency as the abnormal feature data; respectively use the spatio-temporal feature data, the delivery feature data, the preference feature data, the cost feature data, and the abnormal feature data as a feature item to obtain the multiple feature items.

[0117] In a specific application scenario, the obtaining module 301 is configured to query the geographical parameters of the area to be counted, determine the road network information covering the candidate location points from the geographical parameters, and determine the road network level corresponding to the road network information; determine multiple regional buildings in the area to be counted, obtain the building parameters of each regional building in the multiple regional buildings, and determine the building level corresponding to the building parameters of each regional building; calculate the weights of the road network level and the building level, and use the calculated value as the estimated index of the delivery difficulty.

[0118] In a specific application scenario, the obtaining module 301 is configured to perform the following processing on each of the multiple historical orders: query the merchant location information, historical delivery time, and delivery distance of the historical order, calculate the ratio between the delivery distance and the historical delivery time as the delivery speed of the historical order, count the location distance between the merchant location information and the candidate location point, calculate the ratio between the location distance and the delivery speed as the estimated delivery time, and calculate the time difference between the historical delivery time and the estimated delivery time; obtain the multiple time differences of the multiple historical orders, perform logistic regression calculation on the multiple time differences, and use the calculated numerical result as the time estimation index.

[0119] In a specific application scenario, the scoring module 302 is configured to, for each candidate location point, split the location point features of the candidate location point into multiple features to be scored; input the multiple features to be scored into the location point feature model respectively, the location point feature model scores the multiple features to be scored, and performs logistic regression processing on the obtained multiple scores to obtain and output the location point score of the candidate location point; obtain the location point scores of each candidate location point output by the location point feature model to obtain the multiple location point scores.

[0120] In a specific application scenario, the selection module 303 is configured to, when the sorting result indicates that the multiple location point scores are sorted from high to low, select at least one first location point score ranked at the head of the sorting result, determine the target location point corresponding to each first location point score among the multiple candidate location points, and label the corresponding target location point with the at least one first location point score, and push the labeled target location point; when the sorting result indicates that the multiple location point scores are sorted from low to high, select at least one second location point score ranked at the end of the sorting result, determine the target location point corresponding to each second location point score among the multiple candidate location points, and label the corresponding target location point with the at least one second location point score, and push the labeled target location point.

[0121] The device provided by the embodiment of the present application determines multiple candidate location points, obtains the location point features of each candidate location point, including indicators for indicating the environmental state of the candidate location point and indicators for indicating the estimated delivery status after placing the device at the candidate location point, scores the location point features of each candidate location point based on the location point feature model to obtain multiple location point scores, sorts the multiple candidate location points according to the multiple location point scores to obtain a sorting result, and selects a target location point for pushing in the sorting result. By using the location point feature model to fuse and score the multi-dimensional data features of the candidate location points, and comprehensively considering the impact of site selection on the overall delivery scheduling, single-index decision-making is avoided. Not only is the site selection efficiency high, but also the globality, accuracy, and effectiveness of the site selection can be guaranteed.

[0122] It should be noted that for other corresponding descriptions of each functional unit involved in the location selection device provided by the embodiment of the present application, reference can be made to Figure 1 and Figures 2A to 2D the corresponding descriptions therein, which will not be elaborated here.

[0123] In an exemplary embodiment, referring to Figure 4 , a computer device is further provided. The computer device includes a bus, a processor, a memory, and a communication interface, and may further include an input / output interface and a display device. Among them, each functional unit can communicate with each other through the bus. The memory stores a computer program, and the processor is used to execute the program stored on the memory to execute the location selection method in the above embodiment.

[0124] A computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, the steps of the location selection method are implemented.

[0125] Through the description of the above embodiments, those skilled in the art can clearly understand that the present application can be implemented through hardware or by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various implementation scenarios of the present application.

[0126] Those skilled in the art can understand that the drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the drawings are not necessarily essential for implementing the present application.

[0127] Those skilled in the art can understand that the modules in the devices in the implementation scenarios can be distributed in the devices in the implementation scenarios according to the descriptions of the implementation scenarios, or can be correspondingly changed and located in one or more devices different from this implementation scenario. The modules in the above implementation scenarios can be combined into one module, or can be further split into multiple sub-modules.

[0128] The above serial numbers of the present application are only for description and do not represent the advantages or disadvantages of the implementation scenarios.

[0129] The above-disclosed are only several specific implementation scenarios of the present application. However, the present application is not limited thereto, and any changes that can be conceived by those skilled in the art should fall within the protection scope of the present application.

Claims

1. A position selection method, characterized in that, Including: Determine a plurality of candidate location points, and obtain the location point features of each candidate location point among the plurality of candidate location points. The location point features include an index for indicating the environmental state of the candidate location point and an index for indicating the estimated delivery state after installing the device at the candidate location point. Among them, the location point features include spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data. The spatio-temporal feature data includes the historical order quantity within a certain range around the candidate location point and the spatial coverage index of the candidate location point. The delivery feature data includes the estimated delivery difficulty index of the candidate location point and the time estimation index of the candidate location point for delivery scheduling. The abnormal feature data is used to indicate the possible risks in installing the device at the candidate location point. Score the location point features of each candidate location point based on the location point feature model to obtain multiple location point scores of the plurality of candidate location points. The location point feature model is a model trained with the location points of the installed devices as samples for scoring the location point features. Sort the plurality of candidate location points according to the multiple location point scores to obtain a sorting result, and select a target location point from the sorting result for pushing, so that the staff installing the terminal delivery device can refer to the target location point for installing the terminal delivery device.

2. The method according to claim 1, characterized in that, Before determining the plurality of candidate location points and obtaining the location point features of each candidate location point among the plurality of candidate location points, the method further includes: Determine a plurality of installed devices, use the location points where the plurality of installed devices are located as a plurality of sample location points, and use the location point features of the plurality of sample location points as a plurality of sample features. Split each sample feature in the plurality of sample features into a plurality of feature items according to the described feature content, query the feature item scores corresponding to each split feature item, and label each feature item with the feature item scores. Divide the feature items with the same described feature content into the same group to obtain a plurality of feature groups. Train the plurality of feature groups using the logistic regression algorithm to obtain the location point feature model.

3. The method according to claim 1, wherein The determining the plurality of candidate location points and obtaining the location point features of each candidate location point among the plurality of candidate location points includes: Determine the device to be installed and the area to be installed, and obtain the floor area of the device to be installed. Query the area map of the area to be installed, and determine a plurality of candidate areas with an area meeting the floor area of the device on the area map as the plurality of candidate location points. For each candidate location point among the plurality of candidate location points, count the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate location point as a plurality of feature items, and combine the plurality of feature items as the location point features of the candidate location point.

4. The method according to claim 3, wherein The counting the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate location point as a plurality of feature items includes: Taking the candidate location point as the center, an area to be counted with a preset area is circled, the number of historical orders in the area to be counted within a specified historical time period is counted, and the delivery coverage range of the candidate location point is queried. The number of historical orders and the delivery coverage range are used as the spatio-temporal feature data; According to the geographical location parameters of the area to be counted, a delivery difficulty estimation index of the area to be counted is generated, and according to the delivery times of multiple historical orders that occurred in the area to be counted within the specified historical time period, a time estimation index of the candidate location point is determined. The delivery difficulty estimation index and the time estimation index are used as the delivery feature data; Query the resource type of each historical order among the multiple historical orders, calculate the occurrence ratio of each resource type in the multiple historical orders, and obtain the device recognition degree of the device to be placed. The occurrence ratio of each resource type and the device recognition degree are used as the preference feature data; Obtain the device cost data, installation cost data, and maintenance cost data of the device to be placed as the cost feature data; Query the historical placement information of the area to be counted, count the abnormal frequency of abnormalities in the historical placement information, and use the abnormal frequency as the abnormal feature data; Respectively, the spatio-temporal feature data, the delivery feature data, the preference feature data, the cost feature data, and the abnormal feature data are used as a feature item to obtain the multiple feature items.

5. The method according to claim 4, characterized in that, The generating the delivery difficulty estimation index of the area to be counted according to the geographical location parameters of the area to be counted includes: Query the geographical parameters of the area to be counted, determine the road network information covering the candidate location point in the geographical parameters, and determine the road network level corresponding to the road network information; Determine multiple regional buildings in the area to be counted, obtain the building parameters of each regional building in the multiple regional buildings, and determine the building level corresponding to the building parameters of each regional building; Perform a weight calculation on the road network level and the building level, and use the calculated value as the delivery difficulty estimation index.

6. The method according to claim 4, wherein The determining the time estimation index of the candidate location point according to the delivery times of multiple historical orders that occurred in the area to be counted within the specified historical time period includes: Perform the following processing on each historical order among the multiple historical orders: query the merchant location information, historical delivery time, and delivery distance of the historical order, calculate the ratio between the delivery distance and the historical delivery time as the delivery speed of the historical order, count the location distance between the merchant location information and the candidate location point, calculate the ratio between the location distance and the delivery speed as the estimated delivery time, and calculate the time difference between the historical delivery time and the estimated delivery time; Obtain the multiple time differences of the multiple historical orders, perform a logistic regression calculation on the multiple time differences, and use the calculated numerical result as the time estimation index.

7. The method according to claim 1, characterized in that The position feature model scores the position features of each candidate position point to obtain multiple position scores of the multiple candidate position points, including: For each candidate position point, the position feature of the candidate position point is split into multiple feature items to be scored; The multiple feature items to be scored are respectively input into the position feature model. The position feature model scores the multiple feature items to be scored, and performs logistic regression processing on the obtained multiple scores to obtain and output the position score of the candidate position point; The position scores of each candidate position point output by the position feature model are obtained to obtain the multiple position scores.

8. The method according to claim 1, characterized in that Selecting a target position point for pushing in the sorting result includes: When the sorting result indicates that the multiple position scores are sorted from high to low, at least one first position score ranked at the head of the sorting result is selected, the target position point corresponding to each first position score among the multiple candidate position points is determined, and the corresponding target position point is labeled with the at least one first position score, and the labeled target position point is pushed; When the sorting result indicates that the multiple position scores are sorted from low to high, at least one second position score ranked at the end of the sorting result is selected, the target position point corresponding to each second position score among the multiple candidate position points is determined, and the corresponding target position point is labeled with the at least one second position score, and the labeled target position point is pushed.

9. A position selection device, characterized in that, including: An acquisition module, configured to determine multiple candidate position points, and acquire the position features of each candidate position point among the multiple candidate position points. The position features include indicators for indicating the environmental state of the candidate position point and indicators for indicating the estimated delivery state after installing the device at the candidate position point; wherein, the position features include spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data. The spatio-temporal feature data includes the historical order quantity within a certain range around the candidate position point and the spatial coverage index of the candidate position point. The delivery feature data includes the estimated delivery difficulty index of the candidate position point and the time estimation index of the candidate position point for delivery scheduling. The abnormal feature data is used to indicate the possible risks of installing the device at the candidate position point; A scoring module, configured to score the position features of each candidate position point based on a position feature model to obtain multiple position scores of the multiple candidate position points. The position feature model is a model trained with the position points of the installed devices as samples for scoring position features; A selection module, configured to sort the multiple candidate position points according to the multiple position scores to obtain a sorting result, and select a target position point for pushing in the sorting result for the staff installing the end delivery device to refer to the target position point for installing the end delivery device.

10. The device according to claim 9, characterized in that, The device further includes: A determination module, configured to determine a plurality of installed devices, use the location points where the plurality of installed devices are located as a plurality of sample location points, and use the location point features of the plurality of sample location points as a plurality of sample features; A labeling module, configured to split each sample feature in the plurality of sample features into a plurality of feature items according to the described feature content, query the feature item scores corresponding to each split feature item, and label each feature item using the feature item scores; A grouping module, configured to divide the feature items with the same described feature content into the same group to obtain a plurality of feature groups; A training module, configured to train the plurality of feature groups using a logistic regression algorithm to obtain the location point feature model.

11. The device according to claim 9, characterized in that, The obtaining module is configured to determine a device to be installed and an area to be installed, and obtain the floor area of the device to be installed; query the area map of the area to be installed, and determine a plurality of candidate areas on the area map whose areas meet the floor area of the device as the plurality of candidate location points; for each candidate location point in the plurality of candidate location points, count the spatio-temporal feature data, delivery feature data, preference feature data, cost feature data, and abnormal feature data of the candidate location point as a plurality of feature items, and combine the plurality of feature items as the location point feature of the candidate location point.

12. The device according to claim 11, wherein, The obtaining module is configured to circle a to-be-counted area with a preset area centered on the candidate location point, count the number of historical orders in the to-be-counted area during a specified historical time period, and query the delivery coverage range of the candidate location point, and use the number of historical orders and the delivery coverage range as the spatio-temporal feature data; Generate a delivery difficulty estimation index for the to-be-counted area according to the geographical location parameters of the to-be-counted area, and determine a time estimation index for the candidate location point according to the delivery times of a plurality of historical orders that occurred in the to-be-counted area during the specified historical time period, and use the delivery difficulty estimation index and the time estimation index as the delivery feature data; Query the resource types ordered in each of the plurality of historical orders, calculate the occurrence ratio of each resource type in the plurality of historical orders, and obtain the device recognition degree of the device to be installed, and use the occurrence ratio of each resource type and the device recognition degree as the preference feature data; Obtain the device cost data, installation cost data, and maintenance cost data of the device to be installed as the cost feature data; Query the historical installation information of the to-be-counted area, count the abnormal frequency of abnormalities that occurred in the historical installation information, and use the abnormal frequency as the abnormal feature data; respectively use the spatio-temporal feature data, the delivery feature data, the preference feature data, the cost feature data, and the abnormal feature data as a feature item to obtain the plurality of feature items.

13. The device according to claim 12, characterized in that, The obtaining module is configured to query the geographical parameters of the to-be-counted area, determine the road network information covering the candidate location point in the geographical parameters, and determine the road network level corresponding to the road network information; Determine multiple regional buildings in the area to be counted, obtain the building parameters of each regional building among the multiple regional buildings, and determine the building level corresponding to the building parameters of each regional building; Perform a weight calculation on the road network level and the building level, and use the calculated value as the estimated distribution difficulty index.

14. The device according to claim 12, characterized in that, The obtaining module is configured to perform the following processing on each of the multiple historical orders: query the merchant location information, historical delivery time, and delivery distance of the historical order, calculate the ratio between the delivery distance and the historical delivery time as the delivery speed of the historical order, count the location distance between the merchant location information and the candidate location point, calculate the ratio between the location distance and the delivery speed as the estimated delivery time, and calculate the time difference between the historical delivery time and the estimated delivery time; Obtain the multiple time differences of the multiple historical orders, perform a logistic regression calculation on the multiple time differences, and use the calculated numerical result as the time estimation index.

15. The device according to claim 9, characterized in that The scoring module is configured to, for each candidate location point, split the location point features of the candidate location point into multiple features to be scored; input the multiple features to be scored into the location point feature model respectively, and the location point feature model scores the multiple features to be scored, and perform a logistic regression process on the obtained multiple scores to obtain and output the location point score of the candidate location point; obtain the location point scores of each candidate location point output by the location point feature model to obtain the multiple location point scores.

16. The device according to claim 9, characterized in that, The selection module is configured to, when the sorting result indicates that the multiple location point scores are sorted from high to low, select at least one first location point score ranked at the head of the sorting result, determine the target location point corresponding to each first location point score among the multiple candidate location points, and use the at least one first location point score to label the corresponding target location point, and push the labeled target location point; when the sorting result indicates that the multiple location point scores are sorted from low to high, select at least one second location point score ranked at the end of the sorting result, determine the target location point corresponding to each second location point score among the multiple candidate location points, and use the at least one second location point score to label the corresponding target location point, and push the labeled target location point.

17. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images