A customer classification method and device

Abstract

The present invention provides a customer classification method and device, which relates to the financial field. The method comprises: obtaining bank history data of customers to be predicted; determining the potential value of customers to be predicted using a pre-trained LightGBM model and the bank history data of customers to be predicted; and classifying customers according to the determined potential value of customers to be predicted. The present invention determines the potential value of customers to be predicted using a pre-trained LightGBM model and the bank history data of customers to be predicted; and classifies customers according to the determined potential value of customers to be predicted. It can well predict the service potential of customers, provide tiered services to customers, and classify customers with different prediction probabilities into different categories.

Description

Technical Field

[0001] The present invention relates to data processing technology, and in particular to a customer classification method and device. Background Art

[0002] The long-tail customers of a bank refer to a group of bank customers who have smaller assets that they can control, but this group has a large number of customers. Currently, the long-tail customers of a bank account for 85% of all customers. This shows that the long-tail customers contain huge service value.

[0003] At present, this part of customers cannot be effectively maintained and expanded in batches intelligently through traditional account manager channels, and it is difficult to fully invest resources in a customer base of this size. It is necessary to provide targeted services in layers, groups, and batches, and it is urgent to identify the future development potential and various characteristics of customers, so as to achieve the effect of serving each customer differently and release customer value. Summary of the invention

[0004] In view of the defects of customer classification in the prior art, in order to solve at least one technical problem, the present invention provides a customer classification method, comprising:

[0005] Obtain historical bank data of the customer to be predicted;

[0006] Determine the potential value of the customer to be predicted using a pre-trained LightGBM model and the bank's historical data of the customer to be predicted; wherein the pre-trained LightGBM model is determined by model training using historical bank customer data;

[0007] Customers are classified according to their potential values ​​to be predicted.

[0008] In the embodiment of the present invention, the method further includes: using historical data of bank customers to determine a pre-trained LightGBM model; which includes:

[0009] Acquire historical bank customer data of bank customers; perform discrete ordered variable screening and continuous variable screening on the historical bank customer data;

[0010] Perform classification training on the screened historical bank customer data and select data that meets the preset importance threshold as training sample data;

[0011] The pre-trained LightGBM model is determined by performing LightGBM model training using the training sample data.

[0012] In the embodiment of the present invention, the discrete ordered variable screening and continuous variable screening of the historical bank customer data includes:

[0013] Performing discrete ordered variable screening on the historical bank customer data according to a preset Kendall coefficient threshold;

[0014] The historical bank customer data is screened for continuous variables according to a preset Pearson coefficient threshold.

[0015] In the embodiment of the present invention, the classification training of the filtered historical bank customer data to select data that meets a preset importance threshold as training sample data includes:

[0016] The random forest algorithm is used to classify and train the filtered historical bank customer data to determine the feature importance values ​​of the historical bank customer data;

[0017] The training sample data is selected using the preset importance threshold and the determined feature importance values ​​of the historical bank customer data.

[0018] In the embodiment of the present invention, the classification training of the filtered historical bank customer data to select data that meets a preset importance threshold as training sample data also includes:

[0019] Determine whether the amount of positive sample data in the filtered historical bank customer data meets the preset positive sample concentration threshold;

[0020] If it is determined that the preset positive sample concentration threshold is not met, the SMOTE algorithm is used to increase the positive sample concentration.

[0021] In the embodiment of the present invention, the method of determining the potential value of the customer to be predicted by using the pre-trained LightGBM model and the bank data of the customer to be predicted includes:

[0022] Selecting prediction sample data of the bank data of the customer to be predicted based on the determined training sample data;

[0023] The potential value of the customer to be predicted is determined using the determined prediction sample data and the pre-trained LightGBM model.

[0024] At the same time, the present invention also provides a customer classification device, comprising:

[0025] A data acquisition module is used to obtain the historical bank data of the customer to be predicted;

[0026] A potential determination module, which determines the potential value of the customer to be predicted by using a pre-trained LightGBM model and the bank historical data of the customer to be predicted; wherein the pre-trained LightGBM model is determined by model training using historical bank customer data;

[0027] The classification module is used to classify customers according to the determined potential values ​​of the customers to be predicted.

[0028] In an embodiment of the present invention, the device further comprises: a model training module, for determining a pre-trained LightGBM model using historical data of bank customers;

[0029] The model training module includes:

[0030] A sample data acquisition unit, used to acquire historical bank customer data of bank customers;

[0031] A screening unit, used for screening the historical bank customer data for discrete ordered variables and continuous variables;

[0032] A training sample determination unit is used to perform classification training on the screened historical bank customer data and select data that meets a preset importance threshold as training sample data;

[0033] A training unit is used to perform LightGBM model training using the training sample data to determine the pre-trained LightGBM model.

[0034] In an embodiment of the present invention, the training sample determination unit includes:

[0035] A classification unit, used for performing classification training on the screened historical bank customer data using a random forest algorithm to determine the feature importance values ​​of the historical bank customer data;

[0036] The sample selection unit is used to select training sample data using a preset importance threshold and a determined feature importance value of historical bank customer data.

[0037] In the embodiment of the present invention, the training sample determination unit further includes:

[0038] A sample concentration determination unit is used to determine whether the amount of positive sample data in the screened historical bank customer data meets a preset positive sample concentration threshold;

[0039] The concentration enhancement unit is used to enhance the positive sample concentration by using the SMOTE algorithm when determining that the positive sample concentration does not meet the preset positive sample concentration threshold.

[0040] At the same time, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the above method is implemented when the processor executes the computer program.

[0041] At the same time, the present invention also provides a computer-readable storage medium, which stores a computer program for executing the above method.

[0042] The customer classification method and device of the present invention use the pre-trained LightGBM model and the bank history data of the customer to be predicted to determine the potential value of the customer to be predicted; and classify the customer according to the determined potential value of the customer to be predicted. It can well predict the customer service potential, provide layered services to customers, divide customers with different prediction probabilities into different categories, realize differentiated services, and provide more accurate, humane and diversified services to customers of different categories.

[0043] In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0045] Figure 1 A flow chart of the customer classification method provided by the present invention;

[0046] Figure 2 It is a schematic diagram of an embodiment of the present invention;

[0047] Figure 3 A block diagram of a customer classification device provided by the present invention;

[0048] Figure 4 A schematic diagram of an electronic device provided by an embodiment of the present invention;

[0049] Figure 5 It is a schematic block diagram of the system structure of the electronic device 600 according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0051] The present invention provides a customer classification method, such as Figure 1 As shown, it includes:

[0052] Step S101, obtaining the bank history data of the customer to be predicted;

[0053] Step S102, using a pre-trained LightGBM model and the bank historical data of the customer to be predicted to determine the potential value of the customer to be predicted; wherein the pre-trained LightGBM model is determined by model training using historical bank customer data;

[0054] Step S103: classify customers according to the determined customer potential values ​​to be predicted.

[0055] The customer classification method provided by the present invention uses the pre-trained LightGBM model and the bank historical data of the customer to be predicted to determine the potential value of the customer to be predicted, and classifies the customer according to the determined potential value of the customer to be predicted. It can well predict the customer service potential, provide tiered services to customers, divide customers with different prediction probabilities into different categories, realize differentiated services, and provide more accurate, humane and diversified services to customers of different categories.

[0056] The embodiment of the present invention provides a method for predicting the potential of long-tail customers based on the LightGBM algorithm, which mainly includes the following steps:

[0057] Step 1: Get the historical data of N customers from the database, X = {X1, X2, ..., X i}, where 1≤i≤N.

[0058] In this embodiment, obtaining the historical data of N customers from the database refers to obtaining the data after informing the customers of the detailed information related to obtaining the data and obtaining the authorization and approval of the customers.

[0059] Specifically, the customer's historical data includes:

[0060] Basic customer information, such as gender, age, education level, occupation code, etc.; customer asset information, such as credit card spending amount, average daily assets in the past three months, total assets, total loans, etc.;

[0061] Customer historical behavior information, such as credit card spending amount in the past three months, number of credit card spending in the past three months, total outflow amount from debit cards in the past three months, total outflow number from debit cards in the past three months, total inflow amount from debit cards in the past three months, total inflow number from debit cards in the past three months, spending amount from debit cards in the past three months, number of spending from debit cards in the past three months, etc.

[0062] Step 2: Label the historical data of N customers with y and determine the target variable.

[0063] Specifically, in this embodiment, the data date for obtaining customer information is set as June 2021. In this embodiment, long-tail customers are defined as customers whose average daily financial assets in that month are less than 10,000 yuan (inclusive) among individual customers at the end of June 2021. At the same time, in this embodiment, August is set as the marketing month, and potential customers are defined as customers whose monthly average financial assets among given long-tail customers increase to 50,000 yuan or more in any month from September 2021 to November 2021. The obtained customer data is labeled according to the preset rules. A similar labeling method can also be used to select data from other dates. The label of potential customers is set to 1, and the labels of others are set to 0. Then the overall N customer data can be expressed as, X = {X1, X2,…X N ,Y}.

[0064] That is, in the embodiment of the present invention, the historical data of bank customers is used to determine the pre-trained LightGBM model; it includes:

[0065] Obtain historical banking customer data of banking customers;

[0066] Customers are labeled according to preset labeling rules and their historical bank customer data to identify potential customers.

[0067] Step 3: Clean the N customer data.

[0068] In this embodiment, the cleaning of customer data includes: quantifying, merging, deleting, and other processing of discrete feature variables; performing operations such as filling the same type and removing outliers on continuous feature variables; and, in order to consider the diversity and interpretability of variables, deriving variables based on the original variables and performing one-hot encoding.

[0069] One-hot encoding is to expand the values ​​of discrete features to Euclidean space, making the distance calculation between features more reasonable. For example, if the education level is below undergraduate, undergraduate, and above undergraduate, expand one column into three columns, one column marked below undergraduate, one column marked undergraduate, and one column marked above undergraduate. The variables in each column are 1 for hits and 0 for misses. (One column is derived into three columns). One-hot encoding can achieve variable diversity and interpretability.

[0070] Specifically, in the embodiment of the present invention, the processing of discrete variables includes:

[0071] 1. Quantify and group qualitative variables. For example, to quantify gender, use 1 and 2 to represent male and female respectively, and randomly place unknown gender in 1 and 2. Code display;

[0072] Second, group and merge variables that have a large number of groups and are less useful for actual modeling. For example, educational background information can be divided into below undergraduate, undergraduate, and above undergraduate;

[0073] 3. Delete variables that are greatly affected by subjective factors and have many missing values, such as the time of joining the work, annual income, and the type of unit.

[0074] The treatment of continuous variables includes:

[0075] 1. Fill in the missing values ​​of some variables with the same type. For example, the balance of demand deposits is basically similar to the customer's demand assets, so when the customer's demand deposits are missing, the balance of demand assets is used to fill in the missing values.

[0076] 2. Delete outliers in the basic data. Use the candlestick chart to statistically analyze the data distribution of the field and delete the outliers. For example, directly delete the data of customers under 18 and over 80, with personal assets greater than 600,000 yuan, personal consumer loans greater than 8 million yuan, fund and insurance balances greater than 60,000 yuan, account transaction balances greater than 700,000 yuan, and credit card spending greater than 400,000 yuan.

[0077] Step 4: Screen discrete ordered variables based on the Kendall coefficient.

[0078] The overall N customer data can be expressed as: X = {X1, X2, ... X N ,Y}, each customer i can be expressed as 1≤k≤K, K represents the number of customer feature variables. Two discrete ordered features a and b are selected from them. The elements contained in the two are Where 1≤i≤N. The kendall coefficient between these two features can be expressed as:

[0079]

[0080] Among them, N c Represents the number of pairs with the same order, N d Indicates the number of pairs with different sorting.

[0081] N(N-1) represents the number of pairwise combinations of two feature samples.

[0082] In this embodiment, variables whose absolute values ​​of the kendall coefficients are greater than 0.9 are filtered out.

[0083] The same order means: suppose two random variables are X and Y (which can also be regarded as two sets), the number of their elements is N, and the i-th (1<=i<=N) value of the two random variables is X i , Y iIndicates. The corresponding elements in X and Y form an element pair set XY, which contains the elements (X i ,Y i )(1<=i<=N).

[0084] When any two elements (X i ,Y i ) and (X j ,Y j ) is the same as the order, which means when case 1 or 2 occurs; case 1: X i >X j And Y i >Y j , Case 2: X i <X j And Y i <Y j , the two elements are considered to be consistent.

[0085] When case 3 or 4 occurs, the two elements are considered inconsistent (sorted differently). Case 3: X i >X j And Y i <Y j , Case 4: X i <X j And Y i >Y j .

[0086] When situation 5 or 6 occurs (situation 5: X i =X j , Case 6: Y i =Y j ), these two elements are neither consistent nor inconsistent.

[0087] Step 5: Based on the Pearson coefficient, filter the continuous variables. Select two continuous features a and b, then the elements contained in the two are Where 1≤i≤N, the mean of the two feature samples is:

[0088]

[0089]

[0090] The covariance of two features is:

[0091]

[0092] The Pearson coefficient between features can be expressed as:

[0093]

[0094] Among them, A i =C i a , B i =C i b , S A ,S B They represent the standard deviation of the features.

[0095] In this embodiment, variables with an absolute value of the Pearson coefficient greater than 0.9 are filtered out.

[0096] Step 6: For N customers, if the amount of data with label y as 1 (potential customers) is less than 10%, use the SMOTE algorithm to increase the concentration of positive samples.

[0097] The SMOTE algorithm is an improved solution based on the random oversampling algorithm. The basic idea of ​​the SMOTE algorithm is to analyze minority class samples and artificially synthesize new samples based on minority class samples and add them to the data set. The specific algorithm process is as follows:

[0098] (1) For each sample x in the minority class, calculate its distance to all samples in the minority class sample set using the Euclidean distance as the criterion to obtain its k nearest neighbors.

[0099] (2) A sampling ratio is set according to the sample imbalance ratio to determine the sampling rate N. For each minority class sample x, several samples are randomly selected from its k nearest neighbors. Assume that the selected nearest neighbor is o.

[0100] (3) For each randomly selected neighbor o, a new sample is constructed with the original sample according to the formula o(new) = o + rand(0,1)*(xo).

[0101] Step 7: Use the random forest algorithm to train N data samples and return the importance of the features. Select variables with feature importance greater than 0.01.

[0102] The purpose of variable selection is dimensionality reduction. Only independent variables that have a significant impact on the target variable (screened according to feature importance) are selected into the model, thereby increasing the training speed of the model and reducing interference fields.

[0103] That is, in the embodiment of the present invention, in the process of determining the pre-trained LightGBM model using the historical data of the bank, the customer is labeled according to the preset labeling rules and the historical bank customer data of the bank is obtained, and the potential customers are determined to include:

[0104] Discrete ordered variable screening and continuous variable screening of historical bank customer data;

[0105] Perform classification training on the screened historical bank customer data and select sample data that meets the preset importance threshold as training sample data;

[0106] The LightGBM model is determined by using the training sample data for model training.

[0107] Thus, a trained LightGBM model that can be used to predict customer potential value is obtained. Before the potential value prediction is performed, the bank data of the customer to be predicted is cleaned in the aforementioned step three, and then the prediction sample data for customer potential value prediction is selected from the bank data of the customer to be predicted based on the training sample data determined in step seven.

[0108] The potential value of the customer to be predicted is determined by using the determined prediction sample data and the trained LightGBM model.

[0109] Specifically, in the embodiment of the present invention, 70% of the N data are divided into a training set and 30% are divided into a test set for LightGBM model training. Figure 2 As shown, the training result AUC in this embodiment is 0.96.

[0110] For a binary classifier, the output result label (0 or 1) often depends on the output probability and the predetermined probability threshold. For example, the common threshold is 0.5. Samples greater than 0.5 are considered positive samples, and samples less than 0.5 are considered negative samples. If this threshold is increased, the probability of incorrect prediction will decrease, but the probability of correct prediction will also decrease; if this threshold is reduced, the probability of correct prediction will increase, but the probability of incorrect prediction will also increase. In fact, the selection of this threshold also reflects the classification ability of the classifier to a certain extent. Of course, it is hoped that no matter how large the threshold is selected, the classification can be as correct as possible, that is, it is hoped that the classification ability of the classifier is as strong as possible, which can be understood as a robust ability to a certain extent. The ROC curve focuses on two indicators, TPR and FPR. FPR is the point on the horizontal axis, and TPR is the point on the vertical axis. TPR = FPR, the diagonal line, the result predicted as a positive sample is half correct and half wrong, representing the prediction effect of the random classifier. If the ROC curve is below the diagonal line, it means that the classifier is worse than a random classifier. On the contrary, it is better than a random classifier. Of course, we hope that the ROC curve is as convex as possible to the upper left corner. AUC is the area under the ROC curve. Usually, the AUC value is between 0.5 and 1. The larger the AUC value, the better the model effect.

[0111] In the embodiment of the present invention, the customers are classified according to the determined potential values ​​of the customers to be predicted, which specifically includes classifying the customers according to a preset potential value score threshold and the determined potential values ​​of the customers to be predicted.

[0112] Through the above model, we can get the potential value score of each customer from 0 to 1. The higher the score, the greater the potential. Based on the score, customers are divided into different levels. For example, 0.8-1 is a high-score segment with great potential and key service customer groups. We can arrange one-on-one service by a dedicated person. 0.6-0.8 is a medium-score segment. We can provide targeted batch phone or SMS services. For customers below 0.6, the potential is low and we can provide services based on the abundance of resources. In this way, we can achieve the division of different categories of customers.

[0113] Through this patented model, it is possible to well predict customer service potential, provide tiered services to customers, divide customers with different predicted probabilities into different categories, achieve differentiated services, and make customer services in different categories more precise, humane, and diversified.

[0114] At the same time, the present invention also provides a customer classification device, such as Figure 3 As shown, including:

[0115] The data acquisition module 301 is used to acquire the bank history data of the customer to be predicted;

[0116] The potential determination module 302 determines the potential value of the customer to be predicted by using a pre-trained LightGBM model and the bank historical data of the customer to be predicted; wherein the pre-trained LightGBM model is determined by model training using historical bank customer data;

[0117] The classification module 303 is used to classify customers according to the determined potential values ​​of the customers to be predicted.

[0118] Further, if Figure 4 As shown, in the embodiment of the present invention, the device further includes: a model training module 304, which is used to determine a pre-trained LightGBM model using historical data of bank customers;

[0119] The model training module 304 includes:

[0120] A sample data acquisition unit, used to acquire historical bank customer data of bank customers;

[0121] A screening unit, used for screening the historical bank customer data for discrete ordered variables and continuous variables;

[0122] A training sample determination unit is used to perform classification training on the screened historical bank customer data and select data that meets a preset importance threshold as training sample data;

[0123] A training unit is used to perform LightGBM model training using the training sample data to determine the pre-trained LightGBM model.

[0124] In an embodiment of the present invention, the training sample determination unit includes:

[0125] A classification unit, used for performing classification training on the screened historical bank customer data using a random forest algorithm to determine the feature importance values ​​of the historical bank customer data;

[0126] The sample selection unit is used to select training sample data using a preset importance threshold and a determined feature importance value of historical bank customer data.

[0127] In the embodiment of the present invention, the training sample determination unit further includes:

[0128] A sample concentration determination unit is used to determine whether the amount of positive sample data in the screened historical bank customer data meets a preset positive sample concentration threshold;

[0129] The concentration enhancement unit is used to enhance the positive sample concentration by using the SMOTE algorithm when determining that the positive sample concentration does not meet the preset positive sample concentration threshold.

[0130] The customer classification device of the present invention uses the pre-trained LightGBM model and the bank history data of the customer to be predicted to determine the potential value of the customer to be predicted; and classifies the customer according to the determined potential value of the customer to be predicted. It can well predict the customer service potential, provide layered services to customers, divide customers with different prediction probabilities into different categories, realize differentiated services, and provide more accurate, humane and diversified services to customers of different categories.

[0131] The present invention belongs to the field of big data technology and can also be used in the financial field or other fields. The customer classification method and device provided by the present invention can be used for customer classification in the financial field, and can also be used for customer classification in any field other than the financial field. The present disclosure does not limit the application field of the customer classification method and device.

[0132] This embodiment also provides an electronic device, which may be a desktop computer, a tablet computer, a mobile terminal, etc., but this embodiment is not limited thereto. In this embodiment, the electronic device may refer to the embodiments of the aforementioned method and device, the contents of which are incorporated herein, and the repeated parts are not repeated.

[0133] Figure 5 FIG. 6 is a schematic block diagram of a system structure of an electronic device 600 according to an embodiment of the present invention. Figure 5 As shown, the electronic device 600 may include a central processor 100 and a memory 140; the memory 140 is coupled to the central processor 100. It should be noted that this figure is exemplary; other types of structures may also be used to supplement or replace this structure to implement telecommunication functions or other functions.

[0134] In one embodiment, the customer classification function may be integrated into the central processor 100. The central processor 100 may be configured to perform the following control:

[0135] Obtain historical bank data of the customer to be predicted;

[0136] Determine the potential value of the customer to be predicted using the pre-trained LightGBM model and the bank's historical data of the customer to be predicted;

[0137] Customers are classified according to their potential values ​​to be predicted.

[0138] In another embodiment, the customer classification device may be configured separately from the central processor 100. For example, the customer classification device may be configured as a chip connected to the central processor 100, and the customer classification function is implemented under the control of the central processor.

[0139] like Figure 5 As shown, the electronic device 600 may further include: a communication module 110, an input unit 120, an audio processing unit 130, a display 160, and a power supply 170. It is worth noting that the electronic device 600 does not necessarily have to include Figure 5 In addition, the electronic device 600 may also include Figure 5 For components not shown, reference may be made to the prior art.

[0140] like Figure 5 As shown, the central processor 100 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor devices and / or logic devices. The central processor 100 receives inputs and controls the operations of various components of the electronic device 600.

[0141] The memory 140 may be, for example, one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory or other suitable devices. The above-mentioned information related to the failure may be stored, and a program for executing the relevant information may also be stored. The CPU 100 may execute the program stored in the memory 140 to implement information storage or processing.

[0142] The input unit 120 provides input to the CPU 100. The input unit 120 is, for example, a key or a touch input device. The power supply 170 is used to provide power to the electronic device 600. The display 160 is used to display display objects such as images and text. The display may be, for example, an LCD display, but is not limited thereto.

[0143] The memory 140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 140 may also be some other type of device. The memory 140 includes a buffer memory 141 (sometimes referred to as a buffer). The memory 140 may include an application / function storage unit 142, which is used to store application programs and function programs or processes for executing the operation of the electronic device 600 through the central processor 100.

[0144] The memory 140 may also include a data storage unit 143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 144 of the memory 140 may include various drivers for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0145] The communication module 110 is a transmitter / receiver 110 that transmits and receives signals via an antenna 111. The communication module (transmitter / receiver) 110 is coupled to the central processor 100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0146] Based on different communication technologies, multiple communication modules 110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless LAN module. The communication module (transmitter / receiver) 110 is also coupled to a speaker 131 and a microphone 132 via an audio processor 130 to provide an audio output via the speaker 131 and receive an audio input from the microphone 132, thereby realizing a common telecommunication function. The audio processor 130 may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor 130 is also coupled to the central processor 100, so that the sound can be recorded on the local machine through the microphone 132, and the sound stored on the local machine can be played through the speaker 131.

[0147] An embodiment of the present invention further provides a computer-readable program, wherein when the program is executed in an electronic device, the program enables a computer to execute the customer classification method as described in the above embodiment in the electronic device.

[0148] An embodiment of the present invention further provides a storage medium storing a computer-readable program, wherein the computer-readable program enables a computer to execute the customer classification described in the above embodiment in an electronic device.

[0149] The preferred embodiments of the present invention have been described above with reference to the accompanying drawings. Many features and advantages of these embodiments are clear from this detailed description, and therefore the appended claims are intended to cover all such features and advantages of these embodiments that fall within their true spirit and scope. In addition, since many modifications and changes are readily apparent to those skilled in the art, the embodiments of the present invention are not intended to be limited to the precise structure and operation illustrated and described, but all suitable modifications and equivalents that fall within their scope may be included.

[0150] 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.

[0151] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0152] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0153] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0154] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A customer classification method, characterized in that: The method includes: Obtaining historical bank data of the customer to be predicted, wherein the historical bank data includes basic customer information, customer asset information, and customer historical behavior information; Determine the potential value of the customer to be predicted by using a pre-trained LightGBM model and the bank historical data of the customer to be predicted; wherein the pre-trained LightGBM model is determined by using historical bank customer data for model training, and the specific training process includes: obtaining the historical bank customer data of the bank customer; screening the historical bank customer data for discrete ordered variables according to a preset Kendall coefficient threshold, and screening the historical bank customer data for continuous variables according to a preset Pearson coefficient threshold; using a random forest algorithm to perform classification training on the screened historical bank customer data to determine the feature importance value of the historical bank customer data; using the preset importance threshold and the determined feature importance value of the historical bank customer data to select training sample data; using the training sample data to perform LightGBM model training to determine the pre-trained LightGBM model; Customers are classified according to their potential values ​​to be predicted.

2. The customer classification method according to claim 1, characterized in that: After selecting the training sample data, it also includes: Determine whether the amount of positive sample data in the filtered historical bank customer data meets the preset positive sample concentration threshold; If it is determined that the preset positive sample concentration threshold is not met, the SMOTE algorithm is used to increase the positive sample concentration.

3. The customer classification method according to claim 1, characterized in that: The method of using the pre-trained LightGBM model and the bank data of the customer to be predicted to determine the potential value of the customer to be predicted includes: Selecting prediction sample data of the bank data of the customer to be predicted based on the determined training sample data; The potential value of the customer to be predicted is determined using the determined prediction sample data and the pre-trained LightGBM model.

4. A customer classification device, characterized in that: The device comprises: A data acquisition module is used to acquire the bank history data of the customer to be predicted, wherein the bank history data includes the customer's basic information, customer asset information and customer historical behavior information; A potential determination module is used to determine the potential value of the customer to be predicted using a pre-trained LightGBM model and the bank historical data of the customer to be predicted; wherein the pre-trained LightGBM model is determined by model training using historical bank customer data, and the specific training process includes: obtaining the historical bank customer data of the bank customer; screening the historical bank customer data for discrete ordered variables according to a preset Kendall coefficient threshold, and screening the historical bank customer data for continuous variables according to a preset Pearson coefficient threshold; using a random forest algorithm to perform classification training on the screened historical bank customer data to determine the feature importance value of the historical bank customer data; selecting training sample data using the preset importance threshold and the determined feature importance value of the historical bank customer data; and using the training sample data to perform LightGBM model training to determine the pre-trained LightGBM model; The classification module is used to classify customers according to the determined potential values ​​of the customers to be predicted.

5. 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 3 is implemented.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for executing the method according to any one of claims 1 to 3.

Images