Potential customer mining method and device, electronic equipment and storage medium

Abstract

The present application provides a potential customer mining method and device, electronic equipment and storage medium, the method comprises: obtaining the portrait attribute characteristics of the target customer, the portrait attribute characteristics include customer age, customer loyalty, social closeness, shopping APP use active period, online access activity and offline access activity; the portrait attribute characteristics of the target customer are input into the mining model, and the potential tendency degree output by the mining model is obtained, the potential tendency degree is used for representing the probability value of the target user being a new user in the target business circle, the portrait attribute characteristics of the target customer are obtained through the mobile big data of the target customer, the portrait attribute characteristics of the target customer are input into the mining model, the potential tendency degree output by the mining model is obtained, the big data of the target customer for the business circle access is comprehensively supported, and the accuracy of the potential customer prediction is provided.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and in particular to a method, apparatus, electronic device, and storage medium for potential customer mining. Background Technology

[0002] Currently, the rapid development and application of mobile internet, IoT, and 5G communication technologies have led to an explosive growth in data volume and the resulting overload problem. This explosive growth makes it difficult for people to quickly and effectively extract valuable information from the massive and complex datasets. Operators possess ever-increasing amounts of user data, and while bearing high data maintenance costs, how to monetize data assets through technological means and truly generate value from data has become one of the urgent problems they need to solve. With the widespread application of big data technology, businesses are increasingly focusing on potential customer mining, customer needs analysis, targeted lead generation, and precision marketing. Mining potential customers based on mobile big data analytics can help operators monetize data assets and meet businesses' precision marketing needs while reducing marketing costs. However, currently, the only channel for obtaining attribute information needed to build user profiles based on big data is for the client to send a probe request to the terminal to obtain information about the apps downloaded by the client user. This results in insufficiently comprehensive tag attributes when building user profiles, thus affecting the prediction of potential customers. Summary of the Invention

[0003] To address the problems existing in the prior art, the present invention provides a method, apparatus, electronic device, and storage medium for potential customer mining.

[0004] In a first aspect, the present invention provides a method for potential customer mining, comprising:

[0005] Obtain the profile attributes of the target customers, including customer age, customer loyalty, social intimacy, active usage time of shopping apps, online access activity, and offline access activity;

[0006] The profile attributes of the target customer are input into the mining model to obtain the potential tendency output by the mining model. The potential tendency is used to characterize the probability value that the target user is a new user in the target business area.

[0007] The mining model is a model trained by machine learning based on the profile attributes and customer tags of sample users as input, used to predict the potential inclination of users to correspond to a business district.

[0008] The mining model includes a clustering layer and a computation layer. The clustering layer is used to determine the computation cluster corresponding to the target user, and the computation layer is used to determine the potential tendency based on the cluster characteristics of the determined computation cluster and the profile attribute characteristics of the target customer.

[0009] In one embodiment, the step of inputting the target customer's profile attribute features into the mining model to obtain the potential tendency output by the mining model includes:

[0010] The target customer's profile attributes are input into the clustering layer, which then outputs the identifier of the calculated cluster based on the customer's age.

[0011] The identifier of the computing cluster and the profile attribute features of the target customer are input into the computing layer. The computing layer then obtains the profile attribute features of the customer from the corresponding computing cluster based on the identifier of the computing cluster, and outputs the potential tendency based on the obtained profile attribute features of the customer and the profile attribute features of the target customer.

[0012] In one embodiment, obtaining the profile attribute features of the target customer includes:

[0013] Obtain visit records of target customers visiting the target business district within the first statistical period, and determine customer loyalty based on the visit records;

[0014] Obtain the call records of the target customers within the first statistical period, and determine the social intimacy based on the call records;

[0015] Obtain the usage records of shopping apps by target customers within the first statistical period, and determine the active usage periods of shopping apps based on the usage records;

[0016] Obtain the online access records of target customers within the first statistical period, and determine the online access activity based on the access records;

[0017] Obtain visit records of target customers in all business districts within the first statistical period, and determine offline visit activity based on the visit records of all business districts.

[0018] In one embodiment, obtaining the cluster features of the computational cluster includes:

[0019] Obtain the profile attribute features of each sample customer in the computation cluster;

[0020] The age frequency vector is determined based on the number of customers corresponding to each age in the computational cluster.

[0021] The number of customers visiting each business district is determined based on the customer loyalty of each customer in the calculation cluster, and the business district frequency vector is determined based on the number of customers visiting each business district.

[0022] Based on the number of customers using shopping apps during active periods in each customer cluster, determine the active period vector based on the number of customers using shopping apps during active periods;

[0023] The online access activity parameter is determined based on the mean and standard deviation of the online access activity of each customer in the cluster.

[0024] The offline access activity parameters are determined based on the mean and standard deviation of the offline access activity of each customer in the cluster.

[0025] The social intimacy parameter is determined based on the mean and standard deviation of the social intimacy of each customer in the calculation cluster;

[0026] The age frequency vector, business district frequency vector, usage activity parameter, online access activity parameter, offline access activity parameter, and social intimacy parameter are used as cluster features for calculating clusters.

[0027] Secondly, the present invention provides a potential customer mining device, comprising:

[0028] The acquisition module is used to acquire the profile attribute features of the target customer, including customer age, customer loyalty, social intimacy, active time of use of shopping apps, online access activity, and offline access activity.

[0029] The processing module is used to input the profile attribute features of the target customer into the mining model to obtain the potential tendency output by the mining model. The potential tendency is used to characterize the probability value of the target user being a new user in the target business area.

[0030] The mining model is a model trained by machine learning based on the profile attributes and customer tags of sample users as input, used to predict the potential inclination of users to correspond to a business district.

[0031] The mining model includes a clustering layer and a computation layer. The clustering layer is used to determine the computation cluster corresponding to the target user, and the computation layer is used to determine the potential tendency based on the cluster characteristics of the determined computation cluster and the profile attribute characteristics of the target customer.

[0032] Thirdly, the present invention provides an electronic device including a memory and a memory storing a computer program, wherein the processor executes the program to implement the steps of the potential customer mining method described in the first aspect.

[0033] Fourthly, the present invention provides a processor-readable storage medium storing a computer program for causing the processor to perform the steps of the potential customer mining method described in the first aspect.

[0034] The present invention provides a potential customer mining method, apparatus, electronic device and storage medium that obtains profile attribute features of target customers based on mobile big data of target customers' visits to business districts, inputs the profile attribute features of target customers into the mining model, and obtains the potential tendency output by the mining model. It achieves comprehensive support from the big data of customers' visits to business districts and provides accuracy in predicting potential customers. Attached Figure Description

[0035] 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.

[0036] Figure 1 This is a flowchart illustrating the potential customer mining method provided by the present invention;

[0037] Figure 2 This is a schematic diagram of the potential customer discovery device provided by the present invention;

[0038] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention; Detailed Implementation

[0039] 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.

[0040] The following is combined Figures 1-3 The present invention describes a potential customer mining method, apparatus, electronic device, and storage medium.

[0041] Figure 1 A flowchart illustrating a potential customer mining method according to the present invention is shown below. Figure 1 The method includes:

[0042] 11. Obtain the profile attributes of target customers, including customer age, customer loyalty, social intimacy, active time of use of shopping apps, online visit activity, and offline visit activity;

[0043] 12. Input the profile attributes of the target customer into the mining model to obtain the potential propensity output by the mining model. The potential propensity is used to characterize the probability value of the target user being a new user in the target business area.

[0044] Among them, the mining model is a model trained by machine learning based on the profile attributes and customer tags of sample users as input, used to predict the potential inclination of users to correspond to business districts.

[0045] The mining model includes a clustering layer and a computational layer. The clustering layer is used to determine the computational cluster corresponding to the target user, and the computational layer is used to determine the potential tendency based on the cluster characteristics of the determined computational cluster and the profile attribute characteristics of the target customer.

[0046] Regarding steps 11-12, it should be noted that in this invention, the method is used to predict potential customers for business district development. With the rapid development of network information technology and the support of mobile big data, it is possible to analyze data to obtain customer profiles. In the context of the mobile big data era, where customer information is abundant online, customer profiles abstract each specific piece of customer information into tags, using these tags to concretize the customer's image. Based on the customer profile, profile attribute features are extracted. In this invention, profile attribute features include customer age, customer loyalty, social intimacy, active usage times of shopping apps, online activity, and offline activity.

[0047] When customers visit a shopping area using mobile devices, the system stores their movement trajectory. For example, a base station within a shopping area can record the visit records of customers carrying mobile devices to the area.

[0048] When a customer uses a mobile device to make phone calls to others, the system stores that customer's communication history.

[0049] When a customer accesses a shopping app using a mobile device, the system stores their login, browsing, and order history for the shopping app.

[0050] The system preprocesses these data records, including operations such as raw data parsing, data cleaning, data normalization, and formatted data storage, to ensure that the data used later is authentic and accurate.

[0051] By analyzing and refining the preprocessed data, we can determine customer age (registering mobile SIM card numbers using ID card verification), customer loyalty, social intimacy, active periods of shopping app usage, online activity, and offline activity.

[0052] The customer's age is calculated based on the registered mobile SIM card number and verified using an ID card.

[0053] Customer loyalty is calculated based on customers' visit records to the target business area.

[0054] Social intimacy is calculated based on the customer's communication records with other users carrying mobile devices.

[0055] The peak usage periods for shopping apps are calculated based on customers' browsing history of accessing shopping apps using their mobile devices.

[0056] Online visit activity is calculated based on customer order records made using mobile devices.

[0057] Offline visit activity is calculated based on customers' visit records across all business districts.

[0058] The further explanation mainly focuses on the specific processing steps for obtaining the profile attributes of target customers, as detailed below:

[0059] Obtain visit records of target customers to the target business district within the first statistical period, and determine customer loyalty based on the visit records;

[0060] Obtain the call records of target customers within the first statistical period, and determine the social intimacy based on the call records;

[0061] Obtain the usage records of shopping apps by target customers during the first statistical period, and determine the active periods of shopping app usage based on the usage records;

[0062] Obtain the online access records of target customers within the first statistical period, and determine the online access activity based on the access records;

[0063] Obtain visit records of target customers in all business districts within the first statistical period, and determine offline visit activity based on the visit records of all business districts.

[0064] It should be noted that the first statistical period can be one week, one month, or half a year, depending on the specific circumstances.

[0065] Regarding customer loyalty, the visit records it is based on include the total duration and total number of visits to the target business district. The average visit duration can then be calculated, and the logarithm of the average visit duration is used to calculate the customer visit rate.

[0066] For all statistical periods of the target customer, the mean and standard deviation of customer visits and frequency are calculated. Then, for each statistical period, the standard deviation distance of customer visits and frequency relative to the mean is calculated, and scores are obtained based on this distance to obtain a visit score and a visit frequency score. Considering historical records, the loyalty score for the current period is calculated by combining the visit score, visit frequency score, and loyalty score of the previous period. Specifically, the loyalty score for the current period is the sum of the visit score, visit frequency score, and half of the loyalty score of the previous period, where the initial value of the loyalty score for the previous period is 0. Thus, the loyalty score for each period is recursively affected by the loyalty scores of historical periods, with the influence decreasing for older loyalty scores and increasing for more recent loyalty scores.

[0067] For customers who have no visit records for several consecutive periods, a minimum loyalty threshold is set, and customer information below the threshold is temporarily removed to reduce the additional system computing and storage burden caused by the accumulation of lost customers over time.

[0068] Regarding social intimacy, the calculation of call records used to build it includes rankings of average call duration, number of calls, and call density.

[0069] Let's take calculating various rankings of Customer B within Customer A's social circle as an example. First, we calculate the call duration ranking: the average call duration from Customer A to Customer B ranks among all of Customer A's average call durations. Note that this considers off-peak hours and corrects for peak-hour call durations. Second, we calculate the call frequency ranking: the number of times Customer A calls Customer B ranks among all of Customer A's total call frequency. Third, we calculate the call density ranking: for calls between A and B within a month, we sort A's calls to the other end in descending order by the number of days between calls, then in ascending order by the longest interval between calls. Finally, we take the harmonic mean of these three rankings to obtain Customer B's overall social ranking within Customer A's social circle. Similarly, we can calculate Customer A's overall social ranking within Customer B's social circle. The reciprocal of the mean of these two inverse rankings gives the social intimacy level. The calculation process shows that the higher the social intimacy value, the stronger the intimacy between the two parties. The intimacy value is a positive number not exceeding 1, and the intimacy value reaches its maximum value if and only if both parties achieve the highest ranking in each other's social circles.

[0070] The active usage periods for shopping apps are calculated based on user records, including valid app usage data from a customer within a given period. The total traffic value for each of the 24 time slots throughout the day is tallied, and the time slot with the highest total traffic value is identified as the customer's active app usage period. The active usage periods for shopping apps are then calculated based on the total traffic value of that time slot. The shopping apps discussed here refer to mainstream online shopping apps.

[0071] The online visit activity level is based on purchase records including the frequency and intensity of customers' use of online shopping apps (average visit duration).

[0072] First, customer usage across different shopping apps was evaluated separately. Specifically, for each shopping app, the standard deviation of the logarithmic value of the customer's traffic usage relative to the overall mean was calculated. The largest distance among all apps was then taken as the customer's online activity level. The reason for using the maximum value from multiple apps is that, considering user preferences and the inherent differences in traffic generation capabilities among different apps, applying the same standard could easily lead to significant bias. Focusing only on the user's most preferred apps effectively avoids this problem.

[0073] The calculated distances will be corrected for the maximum and minimum values ​​and normalized to a specific interval to reduce the impact of reasonable outliers from individual customers on the overall value.

[0074] Regarding offline visit activity, the data is based on visit records including the frequency and intensity of customers' visits to all offline business districts (average visit duration).

[0075] Offline visit activity includes weekday offline mall visit activity (hereinafter referred to as weekday activity) and weekend offline mall visit activity (hereinafter referred to as weekend activity). Offline visit activity describes the activity level of customers visiting the shopping district and the difference in customer visits between weekdays and weekends. Customers who frequently visit the shopping district are more likely to choose it for convenience and preference, which to some extent reflects their shopping habits in the shopping district.

[0076] In this invention, after obtaining the profile attribute features of the target customer, these features are input into a mining model to obtain the latent tendency score output by the model. The latent tendency score characterizes the probability that the target user is a new user within the target business area. In other words, the mining model analyzes and processes the profile attribute features of the target customer to determine the probability that the target user is likely to be a new customer (potential user). By comparing this probability score with a preset probability threshold, it is possible to predict whether the target user is a potential user within the target business area.

[0077] In this invention, the mining model is a model trained by machine learning based on the profile attributes and customer tags of sample users as input, used to predict the potential inclination of users to correspond to business districts.

[0078] It should be noted that customer tags include new customers and existing customers. New customers are potential customers in the target business district, while existing customers are customers who frequently visit the target business district.

[0079] Here, the profile attributes of each sample user and the customer label to which the sample user belongs are input into the neural network for training to obtain the mining model.

[0080] The trained mining model includes a clustering layer and a computation layer. The clustering layer is used to determine the computation cluster corresponding to the target user, and the computation layer is used to determine the potential tendency based on the cluster features of the determined computation cluster and the profile attribute features of the target customer.

[0081] The clustering layer divides the existing sample customer profile attribute features. In this invention, it divides the customers according to their age and gender, creating clusters for different ages and genders. Each cluster contains the profile attribute features of the customers belonging to that cluster. For example, referencing the overall age distribution of customers, uneven age ranges are divided, resulting in five clusters. Further, each cluster is further split into two clusters based on gender, resulting in a total of ten clusters.

[0082] This computational layer is used to determine potential propensity based on the cluster features of a defined computational cluster and the profile attribute features of the target customer. The clustering layer determines clusters based on the target customer's age. The computational layer then determines the potential propensity based on the cluster features of the defined clusters and the profile attribute features of the target customer.

[0083] The potential customer mining method provided by this invention obtains profile attribute features for business district visits through the mobile big data of target customers, inputs the profile attribute features of target customers into the mining model, and obtains the potential tendency output by the mining model. It is supported by big data of customer visits to business districts to provide accuracy in predicting potential customers.

[0084] The further explanation of the above method mainly focuses on the process of inputting the target customer's profile attribute features into the mining model to obtain the potential tendency output by the mining model, as detailed below:

[0085] The target customer's profile attributes are input into the clustering layer, which then outputs the cluster identifier based on the customer's age.

[0086] The identifier of the computing cluster and the profile attribute features of the target customer are input into the computing layer. The computing layer then retrieves the profile attribute features of the customer from the corresponding computing cluster based on the identifier of the computing cluster, and outputs the potential tendency based on the retrieved profile attribute features of the customer and the profile attribute features of the target customer.

[0087] It should be noted that since clusters are differentiated based on customer age and gender, each cluster has an identifier. The clustering layer can output the identifier of the calculated cluster based on the customer's age or gender. The calculation layer can obtain the customer's profile attribute features from the corresponding calculated cluster based on the identifier of the calculated cluster, and then output the potential tendency based on the obtained customer profile attribute features and the target customer's profile attribute features.

[0088] A further method of the present invention employs a clustering layer and a computation layer to divide the tasks of customer profile attribute features, which can accelerate the processing speed.

[0089] The further explanation of the above method mainly focuses on the process of obtaining the cluster features of the computational cluster, as follows:

[0090] Obtain the profile attribute features of each sample customer in the computation cluster;

[0091] The age frequency vector is determined based on the number of customers corresponding to each age in the computational cluster.

[0092] The number of customers visiting each business district is determined based on the customer loyalty of each customer in the calculation cluster, and the business district frequency vector is determined based on the number of customers visiting each business district.

[0093] Based on the number of customers using shopping apps during active periods in each customer cluster, determine the active period vector based on the number of customers using shopping apps during active periods;

[0094] The online access activity parameter is determined based on the mean and standard deviation of the online access activity of each customer in the cluster.

[0095] The offline access activity parameters are determined based on the mean and standard deviation of the offline access activity of each customer in the cluster.

[0096] The social intimacy parameter is determined based on the mean and standard deviation of the social intimacy of each customer in the calculation cluster;

[0097] The obtained age frequency vector, business district frequency vector, usage activity parameter, online access activity parameter, offline access activity parameter, and social intimacy parameter are used as cluster features for calculating clusters.

[0098] It should be noted that for each cluster, each cluster corresponds to an age range. We can know the number of customers of each age in each sample customer range, and synthesize these customer numbers into a vector to determine the age frequency vector corresponding to each cluster.

[0099] For a business district in a certain region, customer data from several major business districts can be used as data support for the model. To this end, the number of customers visiting different business districts for each sample can be obtained. These customer numbers are then vectorized to determine the business district frequency vector corresponding to each cluster.

[0100] The active usage time vector is determined based on the number of customers using shopping apps during active periods in each cluster.

[0101] The online access activity parameter is calculated by taking one-third of the distance between the mean and the standard deviation of the online access activity for each cluster.

[0102] The offline access activity parameter is calculated by taking one-third of the distance between the mean and the standard deviation of the offline access activity of each cluster. Specifically, it is divided into weekday activity parameter and weekend activity parameter.

[0103] Calculate the mean and standard deviation of social intimacy for all customers within each cluster. The social intimacy parameter is calculated as one-third of the distance between the mean and the standard deviation.

[0104] A further method of the present invention extracts cluster features, which can refine the attribute features of multiple sample users into unified features, reduce the complexity of calculating the potential tendency of target users, and make the potential tendency of target users more accurate.

[0105] The potential customer mining apparatus provided by the present invention is described below. The potential customer mining apparatus described below can be referred to in correspondence with the potential customer mining method described above.

[0106] Figure 2 A schematic diagram of a potential customer discovery device provided by the present invention is shown below. Figure 2 The device includes an acquisition module 21 and a processing module 22, wherein:

[0107] The acquisition module is used to acquire the profile attribute features of the target customer, including customer age, customer loyalty, social intimacy, active time of use of shopping apps, online access activity, and offline access activity.

[0108] The processing module is used to input the profile attribute features of the target customer into the mining model to obtain the potential tendency output by the mining model. The potential tendency is used to characterize the probability value of the target user being a new user in the target business area.

[0109] The mining model is a model trained by machine learning based on the profile attributes and customer tags of sample users as input, used to predict the potential inclination of users to correspond to a business district.

[0110] The mining model includes a clustering layer and a computation layer. The clustering layer is used to determine the computation cluster corresponding to the target user, and the computation layer is used to determine the potential tendency based on the cluster characteristics of the determined computation cluster and the profile attribute characteristics of the target customer.

[0111] In a further description of the above device, the processing module is specifically used for:

[0112] The target customer's profile attributes are input into the clustering layer, which then outputs the identifier of the calculated cluster based on the customer's age.

[0113] The identifier of the computing cluster and the profile attribute features of the target customer are input into the computing layer. The computing layer then obtains the profile attribute features of the customer from the corresponding computing cluster based on the identifier of the computing cluster, and outputs the potential tendency based on the obtained profile attribute features of the customer and the profile attribute features of the target customer.

[0114] In a further description of the above device, the acquisition module is specifically used for:

[0115] Obtain visit records of target customers visiting the target business district within the first statistical period, and determine customer loyalty based on the visit records;

[0116] Obtain the call records of the target customers within the first statistical period, and determine the social intimacy based on the call records;

[0117] Obtain the usage records of shopping apps by target customers within the first statistical period, and determine the active usage periods of shopping apps based on the usage records;

[0118] Obtain the online access records of target customers within the first statistical period, and determine the online access activity based on the access records;

[0119] Obtain visit records of target customers in all business districts within the first statistical period, and determine offline visit activity based on the visit records of all business districts.

[0120] In a further description of the above-described apparatus, the apparatus also includes a construction module, which, in the process of acquiring the cluster characteristics of the computational cluster, is specifically used for:

[0121] Obtain the profile attribute features of each sample customer in the computation cluster;

[0122] The age frequency vector is determined based on the number of customers corresponding to each age in the computational cluster.

[0123] The number of customers visiting each business district is determined based on the customer loyalty of each customer in the calculation cluster, and the business district frequency vector is determined based on the number of customers visiting each business district.

[0124] Based on the number of customers using shopping apps during active periods in each customer cluster, determine the active period vector based on the number of customers using shopping apps during active periods;

[0125] The online access activity parameter is determined based on the mean and standard deviation of the online access activity of each customer in the cluster.

[0126] The offline access activity parameters are determined based on the mean and standard deviation of the offline access activity of each customer in the cluster.

[0127] The social intimacy parameter is determined based on the mean and standard deviation of the social intimacy of each customer in the calculation cluster;

[0128] The age frequency vector, business district frequency vector, usage activity parameter, online access activity parameter, offline access activity parameter, and social intimacy parameter are used as cluster features for calculating clusters.

[0129] Since the device described in this embodiment of the invention is based on the same principle as the method described in the above embodiments, more detailed explanations will not be repeated here.

[0130] It should be noted that, in the embodiments of the present invention, the relevant functional modules can be implemented by a hardware processor.

[0131] The potential customer mining device provided by this invention obtains profile attribute features for business district visits through the mobile big data of target customers, inputs the profile attribute features of target customers into the mining model, and obtains the potential tendency output by the mining model. It is supported by big data of customer visits to business districts to provide accuracy in predicting potential customers.

[0132] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3As shown, the electronic device may include: a processor 31, a communication interface 32, a memory 33, and a communication bus 34, wherein the processor 31, the communication interface 32, and the memory 33 communicate with each other through the communication bus 34. The processor 31 can call the computer program in the memory 33 to execute the steps of the user's potential customer mining method, such as: obtaining the profile attribute features of the target customer, including customer age, customer loyalty, social intimacy, active usage time of shopping apps, online access activity, and offline access activity; inputting the profile attribute features of the target customer into the mining model to obtain the potential propensity output by the mining model, the potential propensity being used to characterize the probability value of the target user being a new user in the target business area; wherein, the mining model is a model trained by machine learning based on the profile attribute features and customer tags of sample users as input, used to predict the potential propensity of users corresponding to business areas; the mining model includes a clustering layer and a computation layer, the clustering layer is used to determine the computation cluster corresponding to the target user, and the computation layer is used to determine the potential propensity based on the cluster features of the determined computation cluster and the profile attribute features of the target customer.

[0133] Furthermore, the logical instructions in the aforementioned memory 33 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, 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.

[0134] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein when the program instructions are executed by a computer, the computer is able to execute the potential customer mining method provided by the above methods, the method comprising: acquiring the profile attribute features of the target customer, the profile attribute features including customer age, customer loyalty, social intimacy, active usage time of shopping apps, online access activity, and offline access activity; inputting the profile attribute features of the target customer into a mining model to obtain the potential propensity output by the mining model, the potential propensity being used to characterize the probability value of the target user being a new user in the target business district; wherein, the mining model is a model trained by machine learning based on the profile attribute features and customer tags of sample users as input, used to predict the potential propensity of users corresponding to business districts;

[0135] The mining model includes a clustering layer and a computational layer. The clustering layer is used to determine the computational cluster corresponding to the target user, and the computational layer is used to determine the potential tendency based on the cluster characteristics of the determined computational cluster and the profile attribute characteristics of the target customer.

[0136] On the other hand, embodiments of this application also provide a processor-readable storage medium storing a computer program. The computer program is used to cause the processor to execute the potential customer mining methods provided in the above embodiments. For example, it includes: acquiring profile attribute features of target customers, including customer age, customer loyalty, social intimacy, active usage time of shopping apps, online access activity, and offline access activity; inputting the profile attribute features of target customers into a mining model to obtain a potential tendency output by the mining model, whereby the potential tendency is used to characterize the probability that a target user is a new user within a target business district; wherein the mining model is a model trained by machine learning based on the profile attribute features and customer tags of sample users as input, used to predict the potential tendency of a user corresponding to a business district.

[0137] The mining model includes a clustering layer and a computational layer. The clustering layer is used to determine the computational cluster corresponding to the target user, and the computational layer is used to determine the potential tendency based on the cluster characteristics of the determined computational cluster and the profile attribute characteristics of the target customer.

[0138] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).

[0139] 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.

[0140] 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.

[0141] 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 identifying potential customers, characterized in that, include: Obtain the profile attributes of the target customer, including customer age, customer loyalty, social intimacy, active time of use of shopping apps, online access activity, and offline access activity; online access activity includes the frequency and intensity of customer use of online shopping apps; The profile attributes of the target customer are input into the mining model to obtain the potential tendency output by the mining model. The potential tendency is used to characterize the probability value that the target user is a new user in the target business area. The mining model is a model trained by machine learning based on the profile attributes and customer tags of sample users as input, used to predict the potential inclination of users to correspond to a business district. The mining model includes a clustering layer and a computation layer. The clustering layer is used to determine the computation cluster corresponding to the target user, and the computation layer is used to determine the potential tendency based on the cluster characteristics of the determined computation cluster and the profile attribute characteristics of the target customer. The process of inputting the target customer's profile attribute features into the mining model to obtain the potential inclination output by the mining model includes: The target customer's profile attributes are input into the clustering layer, which then outputs the identifier of the calculated cluster based on the customer's age. The identifier of the computing cluster and the profile attribute features of the target customer are input into the computing layer. The computing layer obtains the profile attribute features of the customer from the corresponding computing cluster based on the identifier of the computing cluster, and outputs the potential tendency based on the obtained profile attribute features of the customer and the profile attribute features of the target customer. The acquisition of the cluster features of the computational cluster includes: Obtain the profile attribute features of each sample customer in the computation cluster; The age frequency vector is determined based on the number of customers corresponding to each age in the computational cluster. The number of customers visiting each business district is determined based on the customer loyalty of each customer in the calculation cluster, and the business district frequency vector is determined based on the number of customers visiting each business district. Based on the number of customers using shopping apps during active periods in each customer cluster, determine the active period vector based on the number of customers using shopping apps during active periods; The online access activity parameter is determined based on the mean and standard deviation of the online access activity of each customer in the cluster. The offline access activity parameters are determined based on the mean and standard deviation of the offline access activity of each customer in the cluster. The social intimacy parameter is determined based on the mean and standard deviation of the social intimacy of each customer in the calculation cluster; The age frequency vector, business district frequency vector, usage activity parameter, online access activity parameter, offline access activity parameter, and social intimacy parameter are used as cluster features for calculating clusters.

2. The potential customer mining method according to claim 1, characterized in that, The acquisition of the target customer's profile attribute features includes: Obtain visit records of target customers visiting the target business district within the first statistical period, and determine customer loyalty based on the visit records; Obtain the call records of the target customers within the first statistical period, and determine the social intimacy based on the call records; Obtain the usage records of shopping apps by target customers within the first statistical period, and determine the active usage periods of shopping apps based on the usage records; Obtain the online access records of target customers within the first statistical period, and determine the online access activity based on the access records; Obtain visit records of target customers in all business districts within the first statistical period, and determine offline visit activity based on the visit records of all business districts.

3. A potential customer discovery device, characterized in that, include: The acquisition module is used to acquire the profile attributes of the target customer. The profile attributes include customer age, customer loyalty, social intimacy, active time of use of shopping apps, online access activity, and offline access activity; online access activity includes the frequency and intensity of the customer's use of online shopping apps. The processing module is used to input the profile attribute features of the target customer into the mining model to obtain the potential tendency output by the mining model. The potential tendency is used to characterize the probability value of the target user being a new user in the target business area. The mining model is a model trained by machine learning based on the profile attributes and customer tags of sample users as input, used to predict the potential inclination of users to correspond to a business district. The mining model includes a clustering layer and a computation layer. The clustering layer is used to determine the computation cluster corresponding to the target user, and the computation layer is used to determine the potential tendency based on the cluster characteristics of the determined computation cluster and the profile attribute characteristics of the target customer. The processing module is specifically used for: The target customer's profile attributes are input into the clustering layer, which then outputs the identifier of the calculated cluster based on the customer's age. The identifier of the computing cluster and the profile attribute features of the target customer are input into the computing layer. The computing layer obtains the profile attribute features of the customer from the corresponding computing cluster based on the identifier of the computing cluster, and outputs the potential tendency based on the obtained profile attribute features of the customer and the profile attribute features of the target customer. The apparatus further includes a construction module, which, in the process of acquiring the cluster features of the computational cluster, is specifically used for: Obtain the profile attribute features of each sample customer in the computation cluster; The age frequency vector is determined based on the number of customers corresponding to each age in the computational cluster. The number of customers visiting each business district is determined based on the customer loyalty of each customer in the calculation cluster, and the business district frequency vector is determined based on the number of customers visiting each business district. Based on the number of customers using shopping apps during active periods in each customer cluster, determine the active period vector based on the number of customers using shopping apps during active periods; The online access activity parameter is determined based on the mean and standard deviation of the online access activity of each customer in the cluster. The offline access activity parameters are determined based on the mean and standard deviation of the offline access activity of each customer in the cluster. The social intimacy parameter is determined based on the mean and standard deviation of the social intimacy of each customer in the calculation cluster; The age frequency vector, business district frequency vector, usage activity parameter, online access activity parameter, offline access activity parameter, and social intimacy parameter are used as cluster features for calculating clusters.

4. The potential customer discovery device according to claim 3, characterized in that, The acquisition module is specifically used for: Obtain visit records of target customers visiting the target business district within the first statistical period, and determine customer loyalty based on the visit records; Obtain the call records of the target customers within the first statistical period, and determine the social intimacy based on the call records; Obtain the usage records of shopping apps by target customers within the first statistical period, and determine the active usage periods of shopping apps based on the usage records; Obtain the online access records of target customers within the first statistical period, and determine the online access activity based on the access records; Obtain visit records of target customers in all business districts within the first statistical period, and determine offline visit activity based on the visit records of all business districts.

5. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the potential customer mining method according to any one of claims 1 to 2.

6. A processor-readable storage medium, characterized in that, The processor-readable storage medium stores a computer program for causing the processor to perform the steps of the potential customer mining method according to any one of claims 1 to 2.

Images