Target customer determination method and device

Through machine learning, the classification model is trained and the multi-dimensional characteristics of customers is used to solve the problem of private banks accurately identifying high-net-worth target customers from a large number of customers, and efficient target customer mining is achieved.

CN112835958BActive Publication Date: 2024-11-05INDUSTRIAL AND COMMERCIAL BANK OF CHINA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110153204.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-02-04
Publication Date
2024-11-05
Estimated Expiration
2041-02-04

Smart Images

  • Figure CN112835958B_ABST
    Figure CN112835958B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and device for determining a target customer, which can be used in the financial field or other technical fields. The method for determining a target customer includes: obtaining a positive sample customer set and a customer set to be evaluated; training multiple classification models using a preset classification algorithm based on the positive sample customer set and the customer set to be evaluated, and obtaining the predicted classification results of the customer to be evaluated corresponding to each classification model; and determining the target customer in the customer set to be evaluated according to the predicted classification results of the customer to be evaluated corresponding to each classification model. The present invention uses a machine learning model to mine high-net-worth target customers of private banks from a large number of customers, achieving the beneficial effect of accurately and efficiently determining the target customers.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method and device for determining a target customer. Background Art

[0002] Private banks mainly serve high-net-worth individuals. As personal wealth accumulates in society, the number of high-net-worth individuals is gradually increasing. How to dig out private banking target customers from a large number of customers is the main challenge for private banks to achieve smart marketing. Currently, private banks often rely on single-dimensional screening such as assets or subjective human judgment to find target customers. Due to the huge amount of customer data and the large individual differences between customers, it is difficult to accurately identify high-net-worth target customers through manual judgment or single-dimensional screening. Therefore, the existing technology lacks a more efficient and accurate method to determine target customers. Summary of the invention

[0003] In order to solve the technical problems in the above-mentioned background technology, the present invention proposes a method and device for determining target customers.

[0004] In order to achieve the above object, according to one aspect of the present invention, a method for determining a target customer is provided, the method comprising:

[0005] Obtain the positive sample customer set and the customer set to be evaluated;

[0006] A plurality of classification models are trained using a preset classification algorithm based on the positive sample customer set and the customer set to be evaluated, and a predicted classification result of the customer to be evaluated corresponding to each classification model is obtained, wherein, when training each classification model, a preset number of customers are randomly selected from the customer set to be evaluated to form a first customer set, and customers that are not selected in the customer set to be evaluated are determined as a second customer set, and a classification model is trained using a customer feature data set corresponding to the first customer set and a customer feature data set corresponding to the positive sample customer set as training data, and the customers in the second customer set are classified according to the classification model and the customer feature data set corresponding to the second customer set, to obtain a predicted classification result of the customer to be evaluated corresponding to the classification model;

[0007] The target customers in the customer set to be evaluated are determined according to the predicted classification results of the customers to be evaluated corresponding to each classification model.

[0008] Optionally, the target customer determination method further includes:

[0009] The positive sample customer set and a customer feature data set corresponding to the positive sample customer set are generated respectively according to the basic characteristics, residential portrait characteristics and consumption portrait characteristics of each customer in the positive sample customer set and the customer set to be evaluated.

[0010] Optionally, the target customer determination method further includes:

[0011] Obtain the customer's residential address information;

[0012] The residential portrait features of the customer are determined according to the residential address information and a preset luxury house set, wherein the luxury house set includes: the luxury house name and the luxury house boundary longitude and latitude range corresponding to the luxury house name.

[0013] Optionally, the target customer determination method further includes:

[0014] Obtain historical consumption data of customers;

[0015] Determine the historical consumption data corresponding to each consumption type of the customer according to the historical consumption data and a preset consumption type set, wherein the consumption type set includes: a plurality of consumption types and consumption data keywords corresponding to each consumption type;

[0016] The customer's consumption portrait features are determined based on the customer's corresponding historical consumption data for each consumption type, wherein the consumption portrait features include: at least one of: the consumption amount sequence of each consumption type, the number of consumption times of each consumption type, and the consumption proportion of each consumption type.

[0017] Optionally, determining the target customers in the set of customers to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model includes:

[0018] Determine all the prediction classification results corresponding to each customer in the customer set to be evaluated from the prediction classification results of the customers to be evaluated corresponding to each classification model;

[0019] The target customers in the customer set to be evaluated are determined according to all the prediction classification results corresponding to each customer.

[0020] Optionally, the target customer determination method further includes:

[0021] According to the consumption place history data of the target customer and the preset expert knowledge, the service product to be recommended corresponding to the target customer is determined, wherein the expert knowledge includes: a plurality of service products and consumption place keywords corresponding to each service product.

[0022] Optionally, the preset number is equal to the number of customers in the positive sample customer set.

[0023] In order to achieve the above object, according to another aspect of the present invention, a target customer determination device is provided, the device comprising:

[0024] A customer set acquisition unit, used to acquire a positive sample customer set and a customer set to be evaluated;

[0025] A classification prediction unit, used to train multiple classification models using a preset classification algorithm based on the positive sample customer set and the customer set to be evaluated, and obtain the predicted classification results of the customers to be evaluated corresponding to each classification model, wherein, when training each classification model, a preset number of customers are randomly selected from the customer set to be evaluated to form a first customer set, and the customers who are not selected in the customer set to be evaluated are determined as the second customer set, and the classification model is trained using the customer feature data set corresponding to the first customer set and the customer feature data set corresponding to the positive sample customer set as training data, and the customers in the second customer set are classified according to the classification model and the customer feature data set corresponding to the second customer set, so as to obtain the predicted classification results of the customers to be evaluated corresponding to the classification model;

[0026] The target customer determination unit is used to determine the target customers in the customer set to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model.

[0027] In order to achieve the above-mentioned purpose, according to another aspect of the present invention, a computer device is also provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the above-mentioned target customer determination method when executing the computer program.

[0028] In order to achieve the above object, according to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed in a computer processor, the steps in the above target customer determination method are implemented.

[0029] The beneficial effects of the present invention are as follows: the embodiment of the present invention trains a classification model through machine learning to identify target customers in the customer set to be evaluated, thereby achieving the beneficial effect of accurately and efficiently mining high-net-worth target customers of private banks from a large number of customers. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0031] Figure 1 is a flow chart of a method for determining target customers according to an embodiment of the present invention;

[0032] Figure 2 is a flow chart of generating customer residence portrait features according to an embodiment of the present invention;

[0033] Figure 3 is a flow chart of generating consumption profile features according to an embodiment of the present invention;

[0034] Figure 4 is a flow chart of the present invention for determining target customers from a set of customers to be evaluated;

[0035] Figure 5 is a structural block diagram of a target customer determination device according to an embodiment of the present invention;

[0036] Figure 6 It is a schematic diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0038] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0039] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or apparatus.

[0040] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0041] It should be noted that the target customer determination method and device of the present invention can be applied to the financial field, and can also be applied to other technical fields.

[0042] The present invention starts from the actual needs of private banking customer development and the goal of achieving smart marketing. It extracts and processes a large amount of customer basic information data, residential address data, consumption transaction data, product holding data, transfer deposit and withdrawal data, three-party transaction data, etc., to obtain important customer portrait features, and then constructs a machine learning model to achieve accurate mining of high-net-worth target customers of private banks.

[0043] Figure 1 is a flow chart of a method for determining target customers according to an embodiment of the present invention. Figure 1 As shown, the target customer determination method of this embodiment includes steps S101 to S103.

[0044] Step S101, obtaining a positive sample customer set and a customer set to be evaluated.

[0045] In the embodiment of the present invention, the present invention uses the lower limit of high net worth customer assets set by business experts as the screening basis, and selects customers who meet the target customer standards from a large number of customers according to the rules, forming a positive sample customer set P as a positive sample for the subsequent machine learning model, such as selecting customers with an average daily asset of more than 8 million in the past year as positive sample customers. In addition, according to the actual situation, select a customer set U to be evaluated, and use the trained classification model to mine potential high net worth target customers from the customer set to be evaluated. For example, the customer set to be evaluated can select customers with an average daily asset of more than 50,000 and less than 8 million in the past year.

[0046] Step S102 : training multiple classification models using a preset classification algorithm according to the positive sample customer set and the customer set to be evaluated, and obtaining the predicted classification results of the customer to be evaluated corresponding to each classification model.

[0047] In this step, when training each classification model, a preset number of customers are randomly selected from the customer set to be evaluated to form a first customer set, and customers that are not selected from the customer set to be evaluated are determined as a second customer set. A classification model is trained using a customer feature data set corresponding to the first customer set and a customer feature data set corresponding to the positive sample customer set as training data. The customers in the second customer set are classified according to the classification model and the customer feature data set corresponding to the second customer set to obtain a predicted classification result of the customers to be evaluated corresponding to the classification model.

[0048] In one embodiment of the present invention, the preset number is equal to the number of customers in the positive sample customer set.

[0049] In the embodiment of the present invention, this step first obtains the positive sample customer set P and the customer feature data set F corresponding to the positive sample customer set P. p, the customer set U to be evaluated and the customer feature data set F corresponding to the customer set U to be evaluated u Among them, the customer feature data set includes each customer's basic features, residential portrait features, and consumer portrait features. The method for establishing each classification model in this step is as follows.

[0050] Input: positive sample customer set P, customer feature data set F p , the customer set U to be evaluated and the customer feature data set F u .

[0051] Output: Model M t Prediction and classification results S for some customers to be evaluated t .

[0052] Algorithm: First, for the base model M t , take the positive sample customer set P as the positive sample, and its corresponding customer feature data set is F t,p ; Randomly select a customer set with the same number of people as P from the customer set to be evaluated U as the negative sample U t,n (i.e. the first customer set in this step), U t,n The corresponding customer feature data set is F t,n ; and the set of unselected customers in U t,u (i.e., the second customer set in this step) as the test set, U t,u The corresponding customer feature data set is F t,u .

[0053] Then use the preset classification model (for example, XGBOOST) as the base model, F t,p UF t,n As the training data set, the trained model is M t Then use model M t For the unselected customer set F t,u Perform prediction classification and obtain the classification result set S t ={(c,S t,c )|c∈U t,u}, where S t,c For model M t The predicted category of customer c. Thus, we get the predicted classification result set S of the T classification models for the customers to be evaluated = {S t |t∈N and 0≤t <T}。

[0054] Step S103 , determining the target customers in the set of customers to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model.

[0055] In an embodiment of the present invention, this step can first determine all the predicted classification results corresponding to each customer in the customer set to be evaluated from all the predicted classification results for the customers to be evaluated, and then judge whether each customer is a target customer based on all the predicted classification results corresponding to each customer, so as to determine all the target customers in the customer set to be evaluated.

[0056] Figure 4 is a flow chart of the present invention for determining target customers from a set of customers to be evaluated, such as Figure 4 As shown, the above step S103 of determining the target customers in the customer set to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model specifically includes step S401 and step S402.

[0057] Step S401 : determining all the predicted classification results corresponding to each customer in the customer set to be evaluated from the predicted classification results of the customer to be evaluated corresponding to each classification model.

[0058] Step S402: determining target customers in the set of customers to be evaluated according to all the prediction classification results corresponding to each customer.

[0059] In an optional embodiment of the present invention, the predicted classification result is a numerical classification result, wherein 1 represents a prediction as a target customer, and 0 represents a prediction as a non-target customer. The present invention calculates based on all the predicted classification results for the customer to obtain the final predicted result of the customer and determines whether the customer is a target customer. Specifically, this step can sum all the predicted classification results of the customer, and then divide the sum by the total number of all the predicted classification results of the customer to obtain the final predicted result of the customer, and then compare the final predicted result of the customer with a preset threshold. If the final predicted result of the customer is greater than the preset threshold, the customer is determined to be a target customer, otherwise it is a non-target customer. In this way, all target customers in the customer set to be evaluated are determined to form a target customer set.

[0060] In an embodiment of the present invention, the predicted classification results of the customers to be evaluated corresponding to each classification model can be used as a vote for the users to be evaluated. The final voting result of customer c (the proportion of votes for target customers) is calculated using the output results of T classification models. The calculation method is as follows.

[0061]

[0062] The sum function represents the sum of the elements in the calculation set, and the count function represents the number of elements in the calculation set. c It represents the percentage of base models that determine customer c as the target customer in the combined model voting results.

[0063] Then, calculate the final recognition result O of customer c c ,

[0064]

[0065] Among them, S threshold It is a manually set screening threshold, usually 0 threshold ≤1.

[0066] Finally, the target customer set A = {U i |O i =1} is to select the customer set whose combined model judgment result is 1 from the customer set U to be evaluated.

[0067] In one embodiment of the present invention, the target customer determination method of the present invention further includes:

[0068] The positive sample customer set and a customer feature data set corresponding to the positive sample customer set are generated respectively according to the basic characteristics, residential portrait characteristics and consumption portrait characteristics of each customer in the positive sample customer set and the customer set to be evaluated.

[0069] In the embodiment of the present invention, the present invention first determines the basic characteristics, residential profile characteristics and consumption profile characteristics of each customer in the positive sample customer set P and the customer set to be evaluated U, and then generates a customer feature data set F corresponding to the positive sample customer set P. p , the customer feature data set F corresponding to the customer set U to be evaluated u .

[0070] In one embodiment of the present invention, the basic characteristics of a customer may include basic information data, product holding data, transfer deposit and withdrawal data, and three-party transaction data. The present invention can extract the basic information data, product holding data, transfer deposit and withdrawal data, and three-party transaction data of the customer for the positive sample customer set P and the customer set to be evaluated U, and integrate them together. Then, according to the business experience rules, inappropriate customers are eliminated, such as high-risk customers. Finally, the basic characteristics set F1 of the customer is obtained by integration = {f 1,c |c∈(PUU)}, where c represents the customer ID, f 1,c Represents the basic characteristics of customer c.

[0071] Figure 2 is a flow chart of generating customer residence portrait features according to an embodiment of the present invention, such as Figure 2 As shown, the process of generating residential portrait features of a customer in the present invention includes step S201 and step S202.

[0072] Step S201, obtaining the customer's residential address information.

[0073] ​Step S202: Determine the residential portrait features of the customer according to the residential address information and a preset luxury house set, where the luxury house set includes: the name of the luxury house and the longitude and latitude range of the luxury house boundary corresponding to the name of the luxury house.

[0074] In the embodiment of the present invention, whether the residential address information of the customer is a high-end residence (hereinafter referred to as a luxury house) can reflect the social assets of the customer and also reflect the overall asset strength of the customer. This step takes generating the residential portrait features of customer c as an example. First, match the customer's residential address information with the name of the luxury house to obtain the result r1, and then determine whether the longitude and latitude in the customer's residential address information are within the longitude and latitude area of the luxury house to obtain the result r2. Finally, generate the customer's residential portrait r by integrating r1 and r2. c . The specific algorithm is as follows.

[0075] Input: Luxury house set L = {L i |i ∈ N and i < kn}, the corresponding longitude and latitude area set Z of the luxury house boundary, where N is the set of natural numbers and kn is the number of luxury houses; the residential address information S of customer c c , corresponding longitude and latitude (x c , y c ).

[0076] Output: The result r of determining whether the residential address of customer c is a luxury house c .

[0077] Algorithm: First, match the residential address string S c with the luxury house set L to obtain the matching result r1. The calculation method is as follows.

[0078]

[0079] where contain(S c , L i ) means that there exists a certain luxury house name L i , satisfying that the residential address information S c contains the luxury house name L i .

[0080] Then determine whether the longitude and latitude point (x c , y c ) of the customer c's residential address is within the closed area enclosed by the longitude and latitude sequence corresponding to the vertices of a certain luxury house boundary polygon, and the result is denoted as r2. Taking the luxury house L h as an example, calculate whether (x c , y c ) is within its vertex sequence Z h = [(x0, y0), (x1, y1),..., (x n-1 , y n-1)], the result f h .f h The calculation process is as follows:

[0081] (1) Yes And i∈N, calculate the intermediate step cnt i The value is as follows:

[0082]

[0083] (2) Calculation result f h :

[0084]

[0085] Thus, the judgment result of whether the longitude and latitude of the residential address of customer c is within the longitude and latitude fence sequence of kn luxury houses [f0,f1,...,f kn-1 ].

[0086] r2 is calculated as follows:

[0087]

[0088] Finally, calculate the result r of whether the residential address of customer c is a luxury house c :

[0089]

[0090] Thus, we can get the customer residential portrait feature set F2 = {f 2,c |c∈(PUU)}, where f 2,c =r c .

[0091] Figure 3 is a flow chart of generating consumption profile features according to an embodiment of the present invention, such as Figure 3 As shown, the process of generating the consumption portrait features of a customer in the present invention includes steps S301 to S303.

[0092] Step S301, obtaining the customer's historical consumption data.

[0093] Step S302, determining the historical consumption data corresponding to each consumption type of the customer according to the historical consumption data and a preset consumption type set, wherein the consumption type set includes: multiple consumption types and consumption data keywords corresponding to each consumption type.

[0094] Step S303: Determine the consumption portrait features of the customer based on the historical consumption data corresponding to each consumption type of the customer. Among them, the consumption portrait features include at least one of the consumption amount sequence of each consumption type, the consumption frequency of each consumption type, and the consumption proportion of each consumption type.

[0095] In the embodiments of the present invention, the size of the consumption ability and the structural composition of the consumption can reflect the asset strength of the customer to a large extent. According to the divided consumption type set C = {C i |i ∈ N and i < cn}, the number of consumption types is cn, and the type C i contains the category keyword set key i = {k i,d |d ∈ N and d < N k,i}, where N k,i is the number of elements in the keyword set key i . Match the customer consumption historical data with the keywords under each category to obtain the consumption characteristics of each customer under each category, and generate the consumption portrait of the customer. For example, consumption amount, frequency, proportion, etc. Take the consumption amount portrait Fea c of customer c as an example, the calculation method is as follows.

[0096] Input: MS = [(tx_site0, tx_amt0), (tx_site1, tx_amt1),...,(tx_site |MS| , tx_amt |MS| )] is the detailed information of each consumption of customer c, where |MS| represents the number of consumption records MS of customer c, tx_site j is the place of consumption, and tx_amt j is the consumption amount. Key = {key i |i ∈ N and i < cn} is the consumption keyword set for each category.

[0097] Output: The consumption amount sequence Fea c of each type of customer c = [fea0, fea1,..., fea cn-1 .

[0098] Algorithm: First, initialize Fea c = [0, 0,..., 0], with a length of cn. Then, traverse the consumption details of customer c for each type C i to calculate the corresponding fea i , and the calculation method is as follows.

[0099] (1) Calculate each consumption record (tx_site j , tx_amt j) belongs to type C i Amount of tmp j .

[0100]

[0101] Among them, contain(tx_site j ,k i,d ) indicates type C i Keyword collection key i There is k i,d Meet tx_site j Contains k i,d .

[0102] (2) Calculate fea i .

[0103]

[0104] According to the above method, the consumption amount sequence of each type of customer c can be calculated to obtain fea c Similarly, the number of times and proportions of each category of consumption for each customer are calculated, and finally the consumption portrait feature set F3 = {f 3,c |c∈(PUU)}, where f 3,c Satisfy the fea c ∈f 3,c .

[0105] In an embodiment of the present invention, after determining the target customer set A from the customer set U to be evaluated, the present invention further generates a customer profile for each target customer in the target customer set A. Specifically, the present invention can screen the customer features in the customer feature data set of the target customer according to the preset customer feature screening parameters to obtain the customer profile of the target customer, wherein the customer feature screening parameters include: a portrait feature screening set and a business screening feature threshold set, the portrait feature screening set is used to record whether the i-th customer feature in the customer feature data set is retained, and the business screening feature threshold set is used to record the conditions that need to be met for the i-th customer feature in the customer feature data set to be retained.

[0106] In the specific implementation, the present invention uses the customer feature screening parameters set by the business experts, including FC = {(i, FC i )|i∈N and i <N FC}(image feature screening set) and FT={(i,FT i )|FT i ∈R and i∈N and i <N FT}(business screening feature threshold set), where feature FC i∈{0,1} indicates whether the i-th feature is retained, FT i Indicates the threshold value that needs to be met when the i-th eigenvalue is retained, N FC and N FT Respectively represent the number of elements in the set FC and FT. Taking customer c as an example, the final retained portrait information Y c ={f c,i |f c,i ∈f c and FC i =1and f c,i 3FT i}. Calculate and select the feature set of each customer, and finally obtain the portrait information Y of the target customer set A = {Y c |c∈A}.

[0107] In an embodiment of the present invention, after determining the target customer, the method of the present invention further includes:

[0108] According to the consumption place history data of the target customer and the preset expert knowledge, the service product to be recommended corresponding to the target customer is determined, wherein the expert knowledge includes: a plurality of service products and consumption place keywords corresponding to each service product.

[0109] Use the location information in the target customer's consumption history data to match expert knowledge to generate marketing recommendations for products and services that customers need. For example, if a customer often consumes at the airport, airport pick-up services can be recommended. Take the example of generating marketing recommendations for service products that customer C may need, and the calculation method is as follows.

[0110] Input: Customer C's consumption location history data Expert knowledge Where |ME c | is the sequence ME c The number of elements, t i Product Service K i Corresponding keyword sequence The number of elements of .

[0111] Output: Service product SK that customer c can recommend c .

[0112] Algorithm: Set the consumption location tx_site of customer c c,m Sequentially and product services K i The corresponding keywords for each location are site i,j Match, when tx_site c,m Contains the keyword site i,j When i SK is calculated from thisc ={K i |contain(tx_site c,m ,site i,j )}, where contain(tx_site c,m ,site i,j ) indicates that the historical data of customer c’s consumption location ME c There is a consumption site tx_site c,m , Product Service K i Corresponding places The keyword site exists in i,j , satisfying tx_site c,m Include site i,j .

[0113] It can be seen from the above embodiments that the present invention adopts multi-dimensional features and uses machine learning models to mine target customers of private banks. Compared with the previous manual screening of target customers through single-dimensional rules, it not only reduces the workload of target customer screening, but also improves the accuracy of target customer screening.

[0114] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0115] Based on the same inventive concept, an embodiment of the present invention further provides a target customer determination device, which can be used to implement the target customer determination method described in the above embodiment, as described in the following embodiment. Since the principle of solving the problem by the target customer determination device is similar to that of the target customer determination method, the embodiment of the target customer determination device can refer to the embodiment of the target customer determination method, and the repeated parts are not repeated. As used below, the term "unit" or "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiment is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0116] Figure 5 is a structural block diagram of a target customer determination device according to an embodiment of the present invention. Figure 5 As shown, the target customer determination device of the embodiment of the present invention includes: a customer set acquisition unit 1, a classification prediction unit 2 and a target customer determination unit 3.

[0117] The customer set acquisition unit 1 is used to acquire a positive sample customer set and a customer set to be evaluated.

[0118] The classification prediction unit 2 is used to train multiple classification models using a preset classification algorithm based on the positive sample customer set and the customer set to be evaluated, and obtain the predicted classification results of the customers to be evaluated corresponding to each classification model, wherein, when training each classification model, a preset number of customers are randomly selected from the customer set to be evaluated to form a first customer set, and the customers who are not selected in the customer set to be evaluated are determined as the second customer set, and the classification model is trained using the customer feature data set corresponding to the first customer set and the customer feature data set corresponding to the positive sample customer set as training data, and the customers in the second customer set are classified according to the classification model and the customer feature data set corresponding to the second customer set to obtain the predicted classification results of the customers to be evaluated corresponding to the classification model.

[0119] In one embodiment of the present invention, the preset number is equal to the number of customers in the positive sample customer set.

[0120] The target customer determination unit 3 is used to determine the target customers in the customer set to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model.

[0121] In one embodiment of the present invention, the target customer determination device of the present invention further includes:

[0122] A customer feature data set generation unit is used to generate the positive sample customer set and the customer feature data set corresponding to the positive sample customer set according to the basic characteristics, residential portrait characteristics and consumption portrait characteristics of each customer in the positive sample customer set and the customer set to be evaluated.

[0123] In one embodiment of the present invention, the target customer determination device of the present invention further includes:

[0124] A residential address information acquisition unit, used to acquire the residential address information of the customer;

[0125] The residential image feature generating unit is used to determine the residential image features of the customer based on the residential address information and a preset luxury house set, wherein the luxury house set includes: the luxury house name and the luxury house boundary longitude and latitude range corresponding to the luxury house name.

[0126] In one embodiment of the present invention, the target customer determination device of the present invention further includes:

[0127] A historical consumption data acquisition unit, used to acquire the historical consumption data of customers;

[0128] A consumption classification unit, used to determine the historical consumption data corresponding to each consumption type of the customer according to the historical consumption data and a preset consumption type set, wherein the consumption type set includes: a plurality of consumption types and consumption data keywords corresponding to each consumption type;

[0129] A consumption portrait feature generation unit is used to determine the customer's consumption portrait features based on the customer's corresponding historical consumption data for each consumption type, wherein the consumption portrait features include: at least one of: the consumption amount sequence of each consumption type, the number of consumption times of each consumption type, and the consumption proportion of each consumption type.

[0130] In one embodiment of the present invention, the target customer determination unit 3 includes:

[0131] A screening module, used to determine all the predicted classification results corresponding to each customer in the customer set to be evaluated from the predicted classification results of the customer to be evaluated corresponding to each classification model;

[0132] The determination module is used to determine the target customer in the set of customers to be evaluated according to all the prediction classification results corresponding to each customer.

[0133] In one embodiment of the present invention, the target customer determination device of the present invention further includes:

[0134] The service product to be recommended determining unit is used to determine the service product to be recommended corresponding to the target customer based on the target customer's consumption place history data and preset expert knowledge, wherein the expert knowledge includes: multiple service products and consumption place keywords corresponding to each service product.

[0135] In order to achieve the above object, according to another aspect of the present application, a computer device is also provided. Figure 6 As shown, the computer device includes a memory, a processor, a communication interface and a communication bus. A computer program that can be executed on the processor is stored in the memory. When the processor executes the computer program, the steps in the above-mentioned embodiment method are implemented.

[0136] The processor may be a central processing unit (CPU). The processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0137] The memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs, non-transitory computer executable programs and units, such as the corresponding program units in the above method embodiments of the present invention. The processor executes various functional applications of the processor and works data processing by running the non-transitory software programs, instructions and modules stored in the memory, that is, implementing the method in the above method embodiments.

[0138] The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created by the processor, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0139] The one or more units are stored in the memory, and when executed by the processor, the method in the above embodiment is performed.

[0140] The specific details of the above-mentioned computer device can be understood by referring to the corresponding related descriptions and effects in the above-mentioned embodiments, and will not be repeated here.

[0141] In order to achieve the above purpose, according to another aspect of the present application, a computer-readable storage medium is also provided, wherein the computer-readable storage medium stores a computer program, and the computer program implements the steps in the above target customer determination method when executed in a computer processor. Those skilled in the art can understand that the implementation of all or part of the process in the above embodiment method can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and the program can include the process of the embodiment of each method as described above when executed. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated: HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above types of memories.

[0142] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0143] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for determining target customers, characterized in that: include: Obtain the positive sample customer set and the customer set to be evaluated; A plurality of classification models are trained using a preset classification algorithm based on the positive sample customer set and the customer set to be evaluated, and a predicted classification result of the customer to be evaluated corresponding to each classification model is obtained, wherein, when training each classification model, a preset number of customers are randomly selected from the customer set to be evaluated to form a first customer set, and customers not selected in the customer set to be evaluated are determined as a second customer set, and a classification model is trained using a customer feature data set corresponding to the first customer set and a customer feature data set corresponding to the positive sample customer set as training data, and the customers in the second customer set are classified according to the classification model and the customer feature data set corresponding to the second customer set, and a predicted classification result of the customer to be evaluated corresponding to the classification model is obtained, and the customer feature data set includes basic features, residential portrait features, and consumption portrait features of the customer; Determine the target customers in the customer set to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model; The customer's residential profile feature set is obtained through the following steps: Match the residential address information S of customer c c with the luxury house set L to obtain the matching result r1, where the luxury house set L = {L i | i ∈ N and i < kn}, and the corresponding set Z of longitude and latitude regions of the luxury house boundaries, where N is the set of natural numbers and kn is the number of luxury houses; Among them, contain(S c ,L i ) indicates that there is a mansion name L i , satisfying the residential address information S c Contains the name of the mansion L i ; Determine the longitude and latitude of the residential address of customer c (x c ,y c ) is within the closed area enclosed by the longitude and latitude sequence corresponding to the vertex of the border polygon of a luxury house. The result is recorded as r2. Specifically, calculate the longitude and latitude point (x c ,y c ) Is it in a luxury house? h The vertex sequence Z h =[(x0,y0),(x1,y1),...,(x n-1 ,y n-1 )], the result f h , f h The calculation process is as follows: right And i∈N, calculate the intermediate step cnt i The value is as follows: Calculation result f h : Calculate whether the longitude and latitude of the residential address of customer c is within the judgment result of the longitude and latitude fence sequence of kn luxury houses [f0,f1,...,f kn-1 ]; r2 is calculated as follows: Finally, calculate the result r of whether the residential address of customer c is a luxury house c : Thus, we can get the customer's residential portrait feature set F2 = {f 2,c |c∈(PUU)}, where f 2,c =r c , P is the positive sample customer set, and U is the customer set to be evaluated.

2. The target customer determination method according to claim 1, characterized in that: Also includes: The positive sample customer set and a customer feature data set corresponding to the positive sample customer set are generated respectively according to the basic characteristics, residential portrait characteristics and consumption portrait characteristics of each customer in the positive sample customer set and the customer set to be evaluated.

3. The target customer determination method according to claim 2, characterized in that: Also includes: Obtain the customer's residential address information; The residential portrait features of the customer are determined according to the residential address information and a preset luxury house set, wherein the luxury house set includes: the luxury house name and the luxury house boundary longitude and latitude range corresponding to the luxury house name.

4. The target customer determination method according to claim 2, characterized in that: Also includes: Obtain the customer's historical consumption data; Determine the historical consumption data corresponding to each consumption type of the customer according to the historical consumption data and a preset consumption type set, wherein the consumption type set includes: a plurality of consumption types and consumption data keywords corresponding to each consumption type; The customer's consumption portrait features are determined based on the customer's corresponding historical consumption data for each consumption type, wherein the consumption portrait features include: at least one of: the consumption amount sequence of each consumption type, the number of consumption times of each consumption type, and the consumption proportion of each consumption type.

5. The target customer determination method according to claim 1, characterized in that: Determining the target customers in the customer set to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model includes: Determine all the prediction classification results corresponding to each customer in the customer set to be evaluated from the prediction classification results of the customers to be evaluated corresponding to each classification model; The target customers in the customer set to be evaluated are determined according to all the prediction classification results corresponding to each customer.

6. The target customer determination method according to claim 1, characterized in that: Also includes: According to the consumption place history data of the target customer and the preset expert knowledge, the service product to be recommended corresponding to the target customer is determined, wherein the expert knowledge includes: a plurality of service products and consumption place keywords corresponding to each service product.

7. The target customer determination method according to claim 1, characterized in that: The preset number is equal to the number of customers in the positive sample customer set.

8. The target customer determination method according to claim 2, characterized in that: Also includes: The customer features in the customer feature data set of the target customer are screened according to the preset customer feature screening parameters to obtain the customer profile of the target customer.

9. A target customer determination device, characterized in that: include: A customer set acquisition unit, used to acquire a positive sample customer set and a customer set to be evaluated; A classification prediction unit, used to train multiple classification models using a preset classification algorithm based on the positive sample customer set and the customer set to be evaluated, and obtain the predicted classification results of the customers to be evaluated corresponding to each classification model, wherein, when training each classification model, a preset number of customers are randomly selected from the customer set to be evaluated to form a first customer set, and the customers who are not selected in the customer set to be evaluated are determined as the second customer set, and the classification model is trained using the customer feature data set corresponding to the first customer set and the customer feature data set corresponding to the positive sample customer set as training data, and the customers in the second customer set are classified according to the classification model and the customer feature data set corresponding to the second customer set to obtain the predicted classification results of the customers to be evaluated corresponding to the classification model, and the customer feature data set includes the basic characteristics, residential portrait characteristics and consumption portrait characteristics of the customers; A target customer determination unit, configured to determine the target customers in the customer set to be evaluated according to the predicted classification results of the customers to be evaluated corresponding to each classification model; The customer's residential profile feature set is obtained through the following steps: Match the residential address information S of customer c c and the luxury house set L to obtain the matching result r1, where the luxury house set L = {L i | i ∈ N and i < kn}, and the corresponding set Z of longitude and latitude regions of the luxury house boundaries, where N is the set of natural numbers and kn is the number of luxury houses; Among them, contain(S c ,L i ) indicates that there is a mansion name L i , satisfying the residential address information S c Contains the name of the mansion L i ; Determine the longitude and latitude of the residential address of customer c (x c ,y c ) is within the closed area enclosed by the longitude and latitude sequence corresponding to the vertex of the border polygon of a luxury house. The result is recorded as r2. Specifically, calculate the longitude and latitude point (x c ,y c ) Is it in a luxury house? h The vertex sequence Z h =[(x0,y0),(x1,y1),...,(x n-1 ,y n-1 )], the result f h , f h The calculation process is as follows: right And i∈N, calculate the intermediate step cnt i The value is as follows: Calculation result f h : Calculate whether the longitude and latitude of the residential address of customer c is within the judgment result of the longitude and latitude fence sequence of kn luxury houses [f0,f1,...,f kn-1 ]; r2 is calculated as follows: Finally, calculate the result r of whether the residential address of customer c is a luxury house c : Thus, we can get the customer's residential portrait feature set F2 = {f 2,c |c∈(PUU)}, where f 2,c =r c , P is the positive sample customer set, and U is the customer set to be evaluated.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 8 is implemented.

11. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed in a computer processor, the computer program implements the method according to any one of claims 1 to 8.