Target customer identification method and device for business scene, and storage medium

Abstract

The invention discloses a target customer identification method and device for a business scene and a storage medium, and relates to the technical field of data processing, and the method comprises the steps: determining customer potential features and customer adaptability features corresponding to a target business scene in customer information according to the target of the target business scene; inputting the customer potential characteristics into a preset potential prediction model to obtain a potential probability value of the customer; when the potential probability value meets a preset potential condition, determining that the customer is a high-potential customer; inputting the customer adaptability feature corresponding to the high-potential customer into a preset matching degree evaluation model to obtain a adaptability probability value of the high-potential customer; and confirming the high-potential customers of which the adaptability probability values meet a preset adaptability condition as target customers of the target business scene. Through dual-target collaborative screening, a new customer group meeting two project objects of high business potential and high business conformity at the same time is screened out, so that the accuracy of customer identification is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, device and storage medium for identifying target customers in a business scenario. Background Technology

[0002] Currently, the banking industry primarily relies on the manual experience of sales staff to identify the value of new account holders. Staff judge whether a customer meets the target customer requirements for a specific business scenario by observing subjective characteristics. The efficiency and accuracy of this method are influenced by the individual qualities and experience of the sales staff, lacking a scientific and objective screening mechanism. This can easily lead to misjudgments of customer value, potentially missing out on high-potential customers and triggering customer complaints, resulting in poor accuracy in customer identification.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a method, device and storage medium for identifying target customers in a business scenario, aiming to solve the technical problem of how to improve the accuracy of screening customers that meet specific business scenarios.

[0005] To address the aforementioned issues, this application provides a method for identifying target customers in a business scenario, the method comprising:

[0006] Based on the objectives of the target business scenario, determine the customer potential characteristics and customer cooperation characteristics corresponding to the target business scenario from the customer information;

[0007] The customer potential characteristics are input into a preset potential prediction model to obtain the customer's potential probability value;

[0008] When the potential probability value meets the preset potential conditions, the customer is determined to be a high-potential customer;

[0009] The customer cooperation characteristics corresponding to the high-potential customer are input into a preset matching evaluation model to obtain the cooperation probability value of the high-potential customer.

[0010] High-potential customers whose cooperation probability values ​​meet preset cooperation conditions are identified as target customers for the target business scenario.

[0011] In one embodiment, the potential prediction model is a logistic regression model, and the step of inputting the customer potential characteristics into a preset potential prediction model to obtain the customer's potential probability value includes:

[0012] The customer potential characteristics are input into the logistic regression model, and the customer potential characteristics are weighted and summed based on the linear regression formula to obtain a linear score;

[0013] The linear score is transformed using the sigmoid function to obtain the potential probability value of the customer.

[0014] In one embodiment, the matching degree evaluation model is a random forest, and the step of inputting the customer potential characteristics into a preset potential prediction model to obtain the customer's potential probability value includes:

[0015] Starting from the root node of a single tree in the random forest, the customer cooperation feature is matched with the node segmentation rule. When a leaf node is reached, the predicted cooperation value of the single tree is obtained.

[0016] The arithmetic mean of the predicted fitness values ​​of each tree in the random forest is obtained to obtain the fitness probability value.

[0017] In one embodiment, the target customer identification method for the business scenario further includes:

[0018] Based on historical data, obtain the actual participation rate and the actual number of participants corresponding to the first historical potential probability value;

[0019] Filter from the first historical potential probability value the second historical potential probability value corresponding to the actual number of participants being greater than or equal to the preset minimum number of people covered by the business;

[0020] The second historical potential probability value with the highest actual participation rate is determined as the initial potential probability threshold.

[0021] In one embodiment, the target customer identification method for the business scenario further includes:

[0022] Based on historical data, obtain the complaint rate and actual number of participants corresponding to the first historical cooperation probability value;

[0023] Filter from the first historical cooperation probability value the second historical cooperation probability value corresponding to the actual number of participants being greater than or equal to the minimum number of people covered by the service;

[0024] The second historical cooperation probability value with the lowest complaint rate is determined as the initial historical cooperation threshold.

[0025] In one embodiment, the target customer identification method for the business scenario further includes:

[0026] If the actual participation rate corresponding to the initial potential probability threshold is less than or equal to the preset minimum participation rate threshold, then the initial potential probability threshold is reduced based on the preset value until the actual participation rate is greater than the minimum participation rate threshold.

[0027] If the complaint rate corresponding to the initial historical cooperation threshold is greater than the preset maximum complaint rate threshold, the historical cooperation threshold is reduced based on the preset value until the complaint rate is less than or equal to the maximum complaint rate threshold.

[0028] In one embodiment, the target customer identification method for the business scenario further includes:

[0029] The reach response rate at each time point is obtained based on historical data, and the reach response rate is compared with the preset average reach response rate.

[0030] The time points where the reach response rate is greater than the average reach response rate are defined as high response periods;

[0031] When a new customer opens an account during the high-response period, the potential and cooperation level of the new customer are predicted.

[0032] In one embodiment, the step of determining customer potential characteristics and customer cooperation characteristics corresponding to the target business scenario in the customer information based on the target business scenario includes:

[0033] Based on a pre-defined scenario indicator mapping rule library, the target business scenario is decomposed into customer potential dimension and customer cooperation dimension to obtain scenario indicators.

[0034] The customer potential characteristics and customer cooperation characteristics corresponding to the scenario indicators are filtered out from the customer information.

[0035] In addition, to achieve the above objectives, this application also proposes a target customer identification device for a business scenario, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the target customer identification method for the business scenario as described above.

[0036] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the target customer identification method for the business scenario described above.

[0037] This application provides a method for identifying target customers in a business scenario. It breaks down the business scenario needs of new bank account holders into the customer's future business potential and their business fit. A potential prediction model transforms abstract potential characteristics into comparable potential probability values, shifting the qualitative assessment of a customer's high business potential to a quantitative result. A matching degree evaluation model analyzes the degree of matching between customer characteristics and the business scenario, converting the fit into a probability value, thus avoiding resource waste or risk associated with high-potential but low-fit customers. Through this dual-objective collaborative screening, a new customer group that simultaneously meets both the goals of high business potential and high business fit is identified, thereby improving the accuracy and efficiency of customer identification. Attached Figure Description

[0038] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A first flowchart illustrating the target customer identification method for the business scenario of this application;

[0041] Figure 2 A second flowchart illustrating the target customer identification method for the business scenario of this application;

[0042] Figure 3 A third flowchart illustrating the target customer identification method for the business scenario of this application;

[0043] Figure 4 This is a schematic diagram of the hardware operating environment involved in the target customer identification method for the business scenario in this application embodiment.

[0044] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0045] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0046] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0047] To achieve the above objectives, this application proposes a method for identifying target customers in a business scenario. Based on the objectives of the target business scenario, the method determines customer potential characteristics and customer cooperation characteristics corresponding to the target business scenario from customer information. The customer potential characteristics are input into a preset potential prediction model to obtain the customer's potential probability value. When the potential probability value meets a preset potential condition, the customer is identified as a high-potential customer. The customer cooperation characteristics corresponding to the high-potential customer are input into a preset matching degree evaluation model to obtain the high-potential customer's cooperation probability value. The high-potential customer whose cooperation probability value meets the preset cooperation condition is confirmed as the target customer of the target business scenario.

[0048] Currently, the banking industry primarily relies on the manual experience of sales staff to identify the value of new account holders. Staff judge whether a customer meets the target customer requirements for a specific business scenario by observing subjective characteristics. The efficiency and accuracy of this method are influenced by the individual qualities and experience of the sales staff, lacking a scientific and objective screening mechanism. This can easily lead to misjudgments of customer value, potentially missing out on high-potential customers and triggering customer complaints, resulting in poor accuracy in customer identification.

[0049] This application provides a method for identifying target customers in a business scenario. It breaks down the business scenario needs of new bank account holders into the customer's future business potential and their business fit. A potential prediction model transforms abstract potential characteristics into comparable potential probability values, shifting the qualitative assessment of a customer's high business potential to a quantitative result. A matching degree evaluation model analyzes the degree of matching between customer characteristics and the business scenario, converting the fit into a probability value, thus avoiding resource waste or risk associated with high-potential but low-fit customers. Through this dual-objective collaborative screening, a new customer group that simultaneously meets both the goals of high business potential and high business fit is identified, thereby improving the accuracy and efficiency of customer identification.

[0050] It should be noted that the executing entity in this embodiment can be a computing service device with network communication and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or apparatus capable of performing the above functions. The following description uses a target customer identification device in a business scenario as an example to illustrate this embodiment and the subsequent embodiments.

[0051] Based on this, embodiments of this application provide a method for identifying target customers in a business scenario, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the target customer identification method for the business scenario described in this application.

[0052] In this embodiment, the target customer identification method for the business scenario is applied to a target customer identification device for the business scenario, and the method includes steps S10 to S50:

[0053] Step S10: Based on the objectives of the target business scenario, determine the customer potential characteristics and customer cooperation characteristics corresponding to the target business scenario in the customer information.

[0054] In this implementation, a comprehensive feature system covering the entire customer lifecycle was pre-built. This system integrates feature information from multiple sources, including basic information that customers must provide when opening an account, historical behavior data of existing customers accumulated within the bank, and expert rules condensed based on the bank's historical data and industry experience.

[0055] Basic account opening information includes: customer ID, name, ID type and number, account opening date, occupation, contact information, initial deposit amount, and business intentions selected during account opening. Customer historical behavior data includes: account transaction records associated with the customer ID, such as transaction amount, frequency, and channel; operation logs, such as the number of times the wealth management section was clicked; marketing response records, such as whether marketing SMS messages were replied to and whether past activities were participated in; customer service interaction records, such as the number of inquiries and complaint records; and expert rules, including qualitative rules based on historical data and industry-standard features. These three types of data are obtained through preset interfaces, such as the account opening system API and the core system data synchronization channel. Basic account opening information is collected in real time, while historical behavior data and expert rules are updated based on preset time periods. For example, historical behavior data is synchronized daily at midnight, and expert rules are updated monthly. The collected data undergoes format validation, checking field completeness and data type compliance. Data that does not meet requirements is marked as needing cleaning, such as records with missing customer IDs, which are temporarily stored in an exception database. The resulting raw data contains basic account opening information, historical behavior data, and expert rules. A main table is created for basic account opening information, indexed by the customer ID. Historical behavior data is linked to the main table via the customer ID. For expert rule-related data, text rules are first converted into structured data containing customer IDs and rule tags. For example, the rule "Stable Income Customer" is converted into "Customer ID = XXX, Tag = Stable Income," and then linked to the main table via the customer ID. After data cleaning operations such as deduplication and outlier correction, a unified customer data base table is obtained. This table uses the customer ID as a unique index and contains integrated data such as each customer's basic attributes, historical behavior, and rule tags.

[0056] The integrated customer data is transformed into structured features, categorized by customer lifecycle stages such as account opening, active, and mature phases, or by feature types such as basic features, behavioral features, and rule features, forming a systematic feature library. Basic features are extracted from the base data table, and behavioral data is statistically analyzed. Transaction records are transformed into average transaction amount and average monthly transaction frequency over the past X months; operation logs are transformed into login count and longest consecutive login days over the past X days; and expert rule tags are characterized, such as converting the stable income customer tag into a binary feature (stable income = 1 / 0) and the low complaint risk tag into a complaint risk level (low / medium / high).

[0057] The customer profile is categorized along two dimensions: the entire customer lifecycle and key features. The lifecycle dimension includes features for the account opening period (e.g., initial deposit, initial business intention); the growth period (e.g., first-month transaction frequency, app usage frequency); and the maturity period (e.g., average asset balance over the past six months, business penetration rate). The key features dimension includes basic attribute features (e.g., occupation, age); financial capacity features (e.g., deposit amount, asset balance); behavioral preference features (e.g., transaction channel preference, business click records); and rule-derived features (e.g., features related to expert tag conversion). Each feature is accompanied by metadata such as description, data type, update frequency, and associated customer lifecycle stage, forming a feature index and resulting in a comprehensive customer lifecycle feature system.

[0058] Different business scenarios have different objectives. First, break down the scenario objectives into quantifiable indicators related to customer potential and customer cooperation, and then identify the required characteristics based on these indicators.

[0059] In one feasible implementation, step S10 may include steps S11 to S12:

[0060] Step S11: Based on the preset scenario indicator mapping rule library, the target business scenario is decomposed into customer potential dimension and customer cooperation dimension to obtain scenario indicators.

[0061] In this embodiment, the preset scenario indicator mapping rule base stores the association between various business scenarios and corresponding evaluation dimensions and specific indicators. By matching the target business scenario with the rule base, the abstract target is decomposed into quantifiable scenario indicators, providing a clear basis for subsequent feature selection.

[0062] The target business scenario's objective description is matched with scenario types in the scenario indicator mapping rule base. For example, keyword matching might determine that it belongs to a new credit card customer activation promotion scenario. Based on the matched scenario type, corresponding customer potential dimension indicators and customer cooperation dimension indicators are extracted from the rule base. This yields the scenario indicators for the target business scenario. For instance, the potential dimension indicators for a credit card promotion scenario are card application qualification compliance and consumption demand intensity, while the cooperation dimension indicators are marketing response speed and activation intention clarity.

[0063] Step S12: Filter out the customer potential characteristics and customer cooperation characteristics corresponding to the scenario indicators from the customer information.

[0064] In this embodiment, for each customer potential dimension indicator, features reflecting that indicator are filtered from customer information. For example, the credit card application qualification compliance indicator filters for features such as whether there is stable income proof and whether the credit record is good; the consumption demand intensity indicator filters for features such as the total number of transactions in the past 3 months and the proportion of online transactions. Similarly, for each customer cooperation dimension indicator, corresponding features are filtered from customer information. For example, the marketing response speed indicator filters for the response time to credit card application invitations; the activation intention clarity indicator filters for features such as whether the customer actively calls customer service to inquire about the activation process and whether they confirm receipt of activation instructions within the app. This yields a set of customer potential features and a set of customer cooperation features corresponding to the scenario indicators. Based on clear scenario indicators, corresponding features are accurately filtered from customer information, transforming abstract evaluation dimensions into quantifiable and analyzable specific features, laying the foundation for subsequent model analysis.

[0065] Step S20: Input the customer potential characteristics into a preset potential prediction model to obtain the customer's potential probability value.

[0066] Step S30: When the potential probability value meets the preset potential conditions, the customer is determined to be a high-potential customer.

[0067] In this embodiment, the potential prediction model is trained using historical data to predict a customer's future business potential. Input features are quantified to output the customer's potential participation probability, providing a basis for screening high-potential customers. Each customer's potential features are input into the potential prediction model; the model calculates the correlation between the features and potential participation ability, outputting a potential probability value for each customer.

[0068] The core of a potential prediction model is to predict the probability that a customer has the ability to participate in a target business based on customer characteristics. Optionally, the potential prediction model can be a logistic regression model, which linearly combines customer characteristics and uses a sigmoid function to map the output to a probability value between 0 and 1, representing the customer's potential probability of participating in the business. Alternatively, the potential prediction model can be a gradient boosting tree model, which iteratively stacks multiple decision trees to capture the non-linear correlation between features and results, ultimately outputting a probability value. Finally, the potential prediction model can be a neural network model, which uses the non-linear activation of multiple layers of neurons to perform high-dimensional mapping of customer characteristics, capture correlation patterns, and output potential probability values.

[0069] By setting preset potential criteria, such as a potential probability value threshold, customers with the potential to participate are selected from the pool of potential clients, narrowing the scope of subsequent evaluations and ensuring that resources are focused on those with basic potential. The potential probability values ​​of all clients are iterated and compared one by one with the preset potential criteria; clients who meet the criteria are marked as high-potential clients, resulting in a list of high-potential clients containing the IDs of eligible clients and their corresponding potential probability values.

[0070] Step S40: Input the customer cooperation characteristics corresponding to the high-potential customer into a preset matching evaluation model to obtain the cooperation probability value of the high-potential customer.

[0071] Step S50: The high-potential customers whose cooperation probability value is greater than the preset cooperation condition are identified as the target customers of the target business scenario.

[0072] In this embodiment, the matching evaluation model is trained using historical data to analyze customer behavioral characteristics and potential tendencies, assessing their suitability for the business scenario. High-potential customer cooperation characteristics are extracted from customer information and input into the matching evaluation model. The model uses algorithms, such as the sigmoid function mapping, to calculate the correlation between these characteristics and cooperation willingness, outputting a cooperation probability value for each high-potential customer, resulting in a customer cooperation probability value lookup table. The cooperation probability values ​​of high-potential customers in the lookup table are iterated through and compared one by one with preset cooperation conditions. Customers meeting the conditions, such as a cooperation probability greater than or equal to a preset threshold, are identified as target customers for the target business scenario, resulting in the final list of target customers. Through preset cooperation conditions, customers meeting the cooperation willingness criteria are further filtered from the high-potential customers, ultimately yielding target customers who are both capable and willing to participate, supporting business personnel in precise outreach.

[0073] The core of the matching evaluation model is to predict whether high-potential customers are willing to cooperate with business outreach, such as accepting telemarketing or clicking on push links. This requires capturing the correlation between customer behavioral preferences, interaction history, and willingness to cooperate. Optionally, the matching evaluation model can be a logistic regression model, a gradient boosting tree model, or a random forest model, where each tree is trained based on randomly sampled features and data, and the final output probability is the sum of the predictions from all trees.

[0074] In this embodiment, the business scenario requirements of new bank account holders are broken down into the customer's future business potential and the customer's business fit. A potential prediction model transforms abstract potential characteristics into comparable potential probability values, shifting the qualitative assessment of a customer's high business potential to a quantitative result. A matching degree evaluation model analyzes the degree of matching between customer characteristics and business scenarios, converting the fit into probability values ​​to avoid resource waste or risk associated with high-potential but poorly matched customers. Through this dual-objective collaborative screening, a new customer group that simultaneously meets both high business potential and high business fit objectives is identified, thereby improving the accuracy of customer identification and business efficiency.

[0075] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description and will not be repeated hereafter. Based on this, step S20 may include steps S21 to S22:

[0076] Step S21: Input the customer potential characteristics into the logistic regression model, and perform a weighted summation of the customer potential characteristics based on the linear regression formula to obtain a linear score.

[0077] Step S22: The linear score is transformed based on the sigmoid function to obtain the potential probability value of the customer.

[0078] In this embodiment, the logistic regression model is pre-trained. Historical data is divided into training and validation sets according to a certain ratio, for example, an 8:2 ratio. The training set is used to learn the weights, and the validation set is used to evaluate the model. The weight vector w is initialized to 0. For the training set, the predicted probability under the current weights is calculated, the log-likelihood loss is calculated to measure the difference between the prediction and the true label, and the partial derivative of the loss with respect to each weight, i.e., the gradient, is obtained. The weights are updated along the gradient direction: w = w + learning rate * gradient. The above process is repeated until the preset number of iterations is reached, or the loss no longer decreases, to obtain the optimal weights. The model is evaluated using the validation set. If the accuracy is less than a preset accuracy threshold, the learning is adjusted and retrained until the accuracy is greater than or equal to the accuracy threshold, resulting in a trained logistic regression model.

[0079] Customer potential features are standardized and converted into numerical input features. Unordered categorical features, such as occupation type, are one-hot encoded to avoid the model misjudging that numerical size represents priority. Standardization eliminates dimensions: standardized value = (original value - minimum feature value) / (maximum feature value - minimum feature value). For example, for a deposit amount range of 1000-10000 yuan, a new customer deposits 5000 yuan, the standardized value = (5000-1000) / (10000-1000) = 0.44. Missing values ​​are imputed: for example, missing first login duration is filled with 0, and missing occupation type is filled with unknown, encoded separately as [0, 0, 1], resulting in a standardized feature matrix.

[0080] For each new customer, their standardized features are substituted into a linear formula to calculate z = w0 + w1x1 + ... + w8x8, where wi is the weight value and xi is the feature value. For example, if a new customer's features are [2 (age group), 1, 0, 0 (company employee), 0.44 (deposit), 1 (check notification)], the calculated z = 1.56. The z is then transformed into a latent probability value using the sigmoid function, where the sigmoid function is: P = 1 / (1 + e^(-1 / 2)). -z Map the linear score z to the interval [0, 1] and output the probability of a customer participating in the business, i.e., the customer's potential probability value. Batch process all new customers and generate a table mapping customer IDs to potential probability values, containing the ID of each new customer and its corresponding potential probability value.

[0081] Based on the first embodiment of this application, in the third embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description and will not be repeated hereafter. Based on this, step S40 may include steps S41 to S42:

[0082] Step S41: Starting from the root node of a single tree in the random forest, match the customer cooperation feature with the node segmentation rule until the leaf node is reached to obtain the cooperation prediction value of the single tree.

[0083] Step S42: Obtain the arithmetic mean of the predicted fitness values ​​of each tree in the random forest to obtain the fitness probability value.

[0084] In this embodiment, during the training of the random forest model, the cooperation feature set and corresponding cooperation labels of historical high-potential new customers are obtained. Using the historical customer ID as the key, the features and labels of each customer are aligned to obtain the training set. The number of trees and the maximum number of features k per tree are preset. For each tree, m samples are randomly selected with replacement from m historical samples to form a dedicated training subset for that tree. For example, the subset of the first tree contains samples 1, 3, 3, and 5, and the subset of the second tree contains samples 2, 4, 4, and 6. The samples not selected are used to evaluate the performance of the tree. During the training of each tree, k features are randomly selected from n cooperation features, and only the selected features are used for node splitting. The dedicated training subset of the tree and the randomly selected feature set are output. For example, 100 trees correspond to 100 different combinations of samples and features. A single decision tree recursively splits the feature space, dividing the samples into several leaf nodes. Samples within each node have similar cooperation labels. Finally, the proportion of cooperation labels within a node is used as the prediction value to capture the local correlation between features and cooperation. When training a single tree, the entire training subset is placed into the root node as the initial sample set to be segmented. From the root to the leaf, for the current node, the selected k features are traversed, and all possible segmentation thresholds are tried for each feature. The impurity reduction value after segmentation is calculated. For example, using Gini impurity, for samples within a node, if the proportion of samples with a matching label of 1 is p, then the impurity = 2p(1-p); the impurity reduction value = parent node impurity - (left child node impurity * left sample proportion + right child node impurity * right sample proportion). The feature and threshold with the largest impurity reduction are selected for segmentation, dividing the samples into left and right child nodes. The above segmentation process is repeated for each child node until the stopping condition is met, such as the number of samples in the node being less than or equal to a preset number, or the impurity reduction value being less than a preset threshold. The node where segmentation stops is the leaf node, and its output value is the mean of the matching labels of all historical samples within that node, resulting in a trained decision tree containing the segmentation rules from root to leaf. After training all trees, a random forest is formed.

[0085] The original cooperation characteristics of high-potential new customers after potential screening are converted into numerical values ​​according to the rules used during training, ensuring that the feature dimensions are consistent with those used during training, resulting in standardized feature vectors. First, single-tree prediction is performed, starting from the root node. Based on the matching between the new customer feature value and the node splitting rule, the tree is distributed layer by layer downwards. For example, if the root node rule is "SMS open = 1", and the new customer feature "SMS open = 1", it enters the right child node; if the child node rule is "Data completeness ≥ 0.9", and the new customer feature is 0.92, it enters the left child node, and so on, until a leaf node is reached. The output value of the leaf node is extracted as the tree's predicted cooperation value for the user. Second, the arithmetic mean of the predicted values ​​from all trees in the forest is obtained to obtain the final cooperation probability value.

[0086] First, logistic regression is used to quickly identify high-potential groups from all new customers, avoiding the need to process the entire dataset directly with complex random forests, thus saving computational resources and improving efficiency. Then, for these narrowed-down high-potential customers, random forests are used to accurately calculate their cooperation levels, ensuring that customers with high cooperation levels are more precisely targeted, thereby improving outreach conversion rates. This approach avoids both the inefficiency of using random forests to process all new customers and the insufficient accuracy of logistic regression in predicting cooperation levels, making the entire process of potential and cooperation assessment before new customer outreach more efficient and accurate.

[0087] Based on the first embodiment of this application, in the fourth embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Steps A10 to A60 may be included before step S10:

[0088] Step A10: Based on historical data, obtain the actual participation rate and the actual number of participants corresponding to the first historical potential probability value.

[0089] In this embodiment, the customer ID is used as the key to associate the first historical potential probability value with whether the customer actually participated in the business. The first historical potential probability value is grouped according to a preset interval, such as grouping probability values ​​of 0.60-0.65 into one group with an interval of 0.05. For each group, the actual number of participants and the actual participation rate are calculated. The actual number of participants is the number of customers in the group who actually participated in the business, and the actual participation rate is the ratio of the actual number of participants in the group to the total number of customers in the group. A correspondence table containing probability value grouping, actual number of participants, and actual participation rate is obtained.

[0090] The probability value of the first historical cooperation level is associated with whether or not a complaint is filed, and the groups are grouped according to the probability value of the first historical cooperation level at preset intervals. The complaint rate is calculated for each group, which is the ratio of the number of customers who complain in the group to the total number of customers in the group, and is associated with the actual number of participants in the group. A table is obtained that includes probability value groups, actual number of participants, and complaint rate.

[0091] Step A20: Select from the first historical potential probability values ​​a second historical potential probability value that corresponds to the actual number of participants being greater than or equal to the preset minimum number of participants covered by the business.

[0092] Step A30: The second historical potential probability value with the highest actual participation rate is determined as the initial potential probability threshold.

[0093] In this embodiment, all probability value groups in the potential probability and participant count comparison table are traversed, the actual number of participants in each group is extracted, and the actual number of participants is compared with the minimum coverage number of the business. Probability value groups where the actual number of participants is greater than or equal to the preset minimum coverage number of the business are selected. The first historical potential probability value corresponding to these qualified probability value groups is determined as the second historical potential probability value. Under the premise that the coverage target is met, the potential probability value with the highest actual participation rate is selected as the initial threshold. The range of potential probability values ​​that meet the coverage target is selected from historical data to ensure that the subsequently selected threshold can meet the basic business reach requirements and avoid insufficient coverage. Under the constraint of meeting the coverage target, the participation rate is maximized, ensuring that the initial threshold can both meet business volume requirements and accurately select high-potential customers.

[0094] Step A40: Based on historical data, obtain the complaint rate and the actual number of participants corresponding to the first historical cooperation probability value.

[0095] Step A50: Select from the first historical cooperation probability values ​​the second historical cooperation probability value corresponding to the actual number of participants being greater than or equal to the minimum number of people covered by the service.

[0096] Step A60: The second historical cooperation probability value with the lowest complaint rate is determined as the initial historical cooperation threshold.

[0097] In this embodiment, all probability value groups in the cooperation probability and complaint number comparison table are traversed to obtain the actual number of participants in each group. The actual number of participants is compared with the minimum coverage number of the business, and probability value groups where the actual number of participants is greater than or equal to the minimum coverage number of the business are selected. The first historical cooperation probability value corresponding to these qualified probability value groups is determined as the second historical cooperation probability value. The initial cooperation threshold meets the minimum coverage number of the business; otherwise, even if the complaint rate is low, the number of customers selected may be too small to support the business. Under the premise that the coverage number meets the standard, the cooperation probability value with the lowest complaint rate is selected as the initial threshold. This ensures coverage while minimizing the risk of complaints after reaching customers, ensuring that the initial threshold can both meet the business volume requirements and effectively control the risk of reaching customers.

[0098] In one feasible implementation, when the actual participation rate corresponding to the initial potential probability threshold is less than or equal to a preset minimum participation rate threshold, the initial potential probability threshold is reduced based on a preset value until the actual participation rate is greater than the minimum participation rate threshold; when the complaint rate corresponding to the initial historical cooperation threshold is greater than a preset maximum complaint rate threshold, the historical cooperation threshold is reduced based on a preset value until the complaint rate is less than or equal to the maximum complaint rate threshold.

[0099] In this implementation, the actual participation rate corresponding to the initial potential probability threshold is obtained and compared with the minimum participation rate threshold. If the actual participation rate is greater than the minimum participation rate threshold, no adjustment is needed, and the initial threshold is retained. If the actual participation rate is less than or equal to the minimum participation rate threshold, the adjustment process begins, where the step size is reduced to adjust the threshold. The new threshold = initial potential probability threshold - reduction step size. If the actual participation rate of the new threshold is greater than the minimum participation rate threshold, the adjustment stops, and the new threshold is determined as the adjusted potential probability threshold. If the target is still not met, the step size reduction step is repeated until the participation rate meets the target. By gradually reducing the threshold, the actual participation rate reaches the business baseline requirement while ensuring coverage, avoiding the problem of insufficient participation despite highly selective customer screening due to an excessively high initial threshold.

[0100] Obtain the complaint rate corresponding to the initial cooperation threshold and compare it with the highest complaint rate threshold. If the complaint rate is less than or equal to the highest complaint rate threshold, no adjustment is needed, and the initial threshold is retained. Otherwise, proceed with the adjustment process, iteratively adjusting in decreasing steps until the target is met. By gradually reducing the threshold, the complaint rate is controlled within an acceptable range for the business while ensuring coverage, avoiding excessive complaint risk due to an unreasonable initial threshold.

[0101] Optionally, each branch can adjust the cooperation probability value and potential probability value at the smallest operational level of the bank, i.e., the branch level, based on its own historical experience, local factors, and performance. Branch-level threshold adjustment rules are pre-set, allowing for adjustment ranges, step sizes, and mandatory verification conditions, thus controlling risk through boundary checks. Local factors are converted into system-accessible indicators, such as the branch's local customer participation rate over the past three months and the ratio of per capita disposable income in the district / county to the bank's average, and are pre-stored in the branch's local indicator library. Corresponding indicators and adjustment suggestion rules are also provided; for example, if the branch's local participation rate is lower than the bank's average by 10%, it is suggested to lower the potential probability threshold by 0.05 as an auxiliary reference for branch adjustments. A separate threshold adjustment backend interface is configured for each branch, supporting input of adjustment values, viewing of local indicators, and submission of adjustment requests; a real-time threshold update interface is set up for the model to call new thresholds. Branch staff input adjustment values ​​through the adjustment operation interface. After the branch submits the adjustment, it is verified whether it complies with the branch-level threshold adjustment rules.

[0102] Based on the first embodiment of this application, in the fifth embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 The target customer identification method for the above business scenarios may also include steps S01 to S03:

[0103] Step S01: Obtain the reach response rate at each time point based on historical data, and compare the reach response rate with the preset average reach response rate.

[0104] Step S02: The time points where the reach response rate is greater than the average reach response rate are determined as high response periods.

[0105] Step S03: When the new customer's account opening time falls within the high response period, perform potential prediction and cooperation prediction for the new customer.

[0106] It's important to clarify that being reached refers to a target audience (such as a customer) receiving information, notifications, or services from a company, organization, or individual. In a banking service scenario, when a bank sends a text message to a newly opened account customer informing them of new customer benefits, the customer is considered to have been reached when they see this message.

[0107] In this embodiment, based on historical new customer reach data, the correlation between time points and response rates is determined. By comparing the averages, the high response periods when customers are more willing to respond are identified. Potential and cooperation predictions are made during the high response periods, which can more accurately screen reach targets and avoid making ineffective predictions and outreach during periods when customers have low attention.

[0108] Historical data is grouped according to preset time periods, for example, by hourly nodes, new customers reached between 09:00 and 10:00 are grouped together. The reach response rate for each group is calculated, which is the ratio of the number of new customers responding within that node to the total number of new customers reached within that node, resulting in a table showing the correlation between time nodes and response rates. The arithmetic mean of the response rates for all time nodes is calculated. The table is then iterated through, and nodes with response rates greater than the mean are marked as high-response periods. Consecutive high-response nodes are merged to form a list of high-response periods.

[0109] Convert the new customer's account opening time to its corresponding hourly node; for example, 09:45 belongs to the 09:00-10:00 node. Query the high-response time period list to determine if the account opening time's node is within the list. If it is, trigger the prediction process.

[0110] In one feasible implementation, a time-period characteristic is added to new customers, such as an account opening during a high-response period (=1). This characteristic serves as input to both the potential prediction model and the cooperation model. The time-period characteristic, along with real-time behavior and basic information, is input into the potential prediction model to calculate a potential probability value. A preset potential threshold is then used to determine if the customer is a high-potential customer. For high-potential customers, the cooperation characteristic is extracted and input into the matching evaluation model to calculate a cooperation probability value. A preset cooperation threshold is then used to determine if the customer is a target customer. Since the high-response period reflects the level of customer interaction willingness, even with limited data available in the initial account opening phase, combining this time-period characteristic with basic account opening characteristics allows for a more accurate assessment of their business participation potential, making the prediction results more reliable.

[0111] This application provides a target customer identification device for a business scenario, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the target customer identification method for the business scenario described in Embodiment 1 above.

[0112] The following is for reference. Figure 4 The diagram illustrates a structural schematic of a target customer identification device suitable for implementing the business scenarios of the embodiments of this application. The target customer identification device in the business scenarios of the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, personal digital assistants (PDAs), tablet computers (PADs), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 4 The target customer identification device shown in the business scenario is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0113] like Figure 4As shown, the target customer identification device for a business scenario may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the target customer identification device for the business scenario. The processing unit 1001, the read-only memory 1002, and the RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the target customer identification device in a business scenario to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a target customer identification device in a business scenario with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0114] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0115] The target customer identification device for business scenarios provided in this application, employing the target customer identification method for business scenarios described in the above embodiments, can solve the technical problem of how to improve the accuracy of screening customers who meet specific business scenarios. Compared with the prior art, the beneficial effects of the target customer identification device for business scenarios provided in this application are the same as those of the target customer identification method for business scenarios provided in the above embodiments, and other technical features in this target customer identification device for business scenarios are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0116] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0117] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0118] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the target customer identification method for the business scenario described in the above embodiments.

[0119] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, radio frequency (RF), etc., or any suitable combination thereof.

[0120] The aforementioned computer-readable storage medium may be included in a target customer identification device for a business scenario; or it may exist independently and not assembled into a target customer identification device for a business scenario. The aforementioned computer-readable storage medium carries one or more programs. When the aforementioned one or more programs are executed by the target customer identification device for a business scenario, the target customer identification device for the business scenario: determines, based on the target of the target business scenario, customer potential characteristics and customer cooperation characteristics corresponding to the target business scenario from customer information; inputs the customer potential characteristics into a preset potential prediction model to obtain a potential probability value for the customer; when the potential probability value meets a preset potential condition, determines the customer as a high-potential customer; inputs the customer cooperation characteristics corresponding to the high-potential customer into a preset matching degree evaluation model to obtain a cooperation probability value for the high-potential customer; and confirms the high-potential customer whose cooperation probability value meets the preset cooperation condition as a target customer of the target business scenario.

[0121] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the client computer, partially on the client computer, as a standalone software package, partially on the client computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the client computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0122] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0123] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0124] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the target customer identification method for the above-described business scenario. This addresses the technical problem of improving the accuracy of screening customers who meet specific business scenarios. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the target customer identification method for the business scenario provided in the above embodiments, and will not be elaborated upon here.

[0125] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for identifying target customers in a business scenario, characterized in that, The target customer identification method for the aforementioned business scenario includes: Based on the objectives of the target business scenario, determine the customer potential characteristics and customer cooperation characteristics corresponding to the target business scenario from the customer information; The customer potential characteristics are input into a preset potential prediction model to obtain the customer's potential probability value; When the potential probability value meets the preset potential conditions, the customer is determined to be a high-potential customer; The customer cooperation characteristics corresponding to the high-potential customer are input into a preset matching evaluation model to obtain the cooperation probability value of the high-potential customer. High-potential customers whose cooperation probability values ​​meet preset cooperation conditions are identified as target customers for the target business scenario.

2. The target customer identification method for a business scenario as described in claim 1, characterized in that, The potential prediction model is a logistic regression model. The step of inputting the customer potential characteristics into the preset potential prediction model to obtain the customer's potential probability value includes: The customer potential characteristics are input into the logistic regression model, and the customer potential characteristics are weighted and summed based on the linear regression formula to obtain a linear score; The linear score is transformed using the sigmoid function to obtain the potential probability value of the customer.

3. The target customer identification method for a business scenario as described in claim 1, characterized in that, The matching degree evaluation model is a random forest, and the step of inputting the customer potential characteristics into a preset potential prediction model to obtain the customer's potential probability value includes: Starting from the root node of a single tree in the random forest, the customer cooperation feature is matched with the node segmentation rule. When a leaf node is reached, the predicted cooperation value of the single tree is obtained. The arithmetic mean of the predicted fitness values ​​of each tree in the random forest is obtained to obtain the fitness probability value.

4. The target customer identification method for a business scenario as described in claim 1, characterized in that, The target customer identification method for the aforementioned business scenario also includes: Based on historical data, obtain the actual participation rate and the actual number of participants corresponding to the first historical potential probability value; Filter from the first historical potential probability value the second historical potential probability value corresponding to the actual number of participants being greater than or equal to the preset minimum number of people covered by the business; The second historical potential probability value with the highest actual participation rate is determined as the initial potential probability threshold.

5. The target customer identification method for a business scenario as described in claim 4, characterized in that, The target customer identification method for the aforementioned business scenario also includes: Based on historical data, obtain the complaint rate and actual number of participants corresponding to the first historical cooperation probability value; Filter from the first historical cooperation probability value the second historical cooperation probability value corresponding to the actual number of participants being greater than or equal to the minimum number of people covered by the service; The second historical cooperation probability value with the lowest complaint rate is determined as the initial historical cooperation threshold.

6. The target customer identification method for a business scenario as described in claim 5, characterized in that, The target customer identification method for the aforementioned business scenario also includes: If the actual participation rate corresponding to the initial potential probability threshold is less than or equal to the preset minimum participation rate threshold, then the initial potential probability threshold is reduced based on the preset value until the actual participation rate is greater than the minimum participation rate threshold. If the complaint rate corresponding to the initial historical cooperation threshold is greater than the preset maximum complaint rate threshold, the historical cooperation threshold is reduced based on the preset value until the complaint rate is less than or equal to the maximum complaint rate threshold.

7. The target customer identification method for a business scenario as described in claim 1, characterized in that, The target customer identification method for the aforementioned business scenario also includes: The reach response rate at each time point is obtained based on historical data, and the reach response rate is compared with the preset average reach response rate. The time points where the reach response rate is greater than the average reach response rate are defined as high response periods; When a new customer opens an account during the high-response period, the potential and cooperation level of the new customer are predicted.

8. The target customer identification method for a business scenario as described in claim 1, characterized in that, The step of determining the customer potential characteristics and customer cooperation characteristics corresponding to the target business scenario from the customer information based on the target business scenario includes: Based on a pre-defined scenario indicator mapping rule library, the target business scenario is decomposed into customer potential dimension and customer cooperation dimension to obtain scenario indicators. The customer potential characteristics and customer cooperation characteristics corresponding to the scenario indicators are filtered out from the customer information.

9. A target customer identification device for a business scenario, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the target customer identification method for any of claims 1 to 8.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the target customer identification method for the business scenario as described in any one of claims 1 to 8.

Images