Sales strategy generation method and device, program product and storage medium

Abstract

Embodiments of the invention provide a sales strategy generation method and apparatus, a program product and a storage medium. The method comprises the steps of obtaining object data and a target transaction factor of a target object; inputting the object data and the target transaction factor into a target clustering model to obtain a clustering feature table of the target object, the clustering feature table comprising category features and category levels of the target object, the target clustering model is a model obtained by performing unsupervised training on the initial clustering model by using a group of sample object data and a group of sample transaction factors; inputting the clustering feature table into a target regression model to obtain a target feature portrait, the target feature portrait being used for representing feature distribution of the target object, and the target regression model being a model determined by performing regression analysis modeling on the sample clustering feature table; and generating a target sales strategy based on the target feature portrait. The problem that the sales strategy cannot be accurately generated in the related technology is solved, and the effect of improving the pertinence of the sales strategy is achieved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technology, and more specifically, to a method and device for generating a sales strategy, a program product, and a storage medium. Background Art

[0002] At present, the sales of financial products are mostly centered on financial products, and often only focus on their own product lines. There is a lack of insight and integration into market trends, resulting in serious information islands. In addition, during the sales process, the analysis of customer needs is not in-depth enough, and the financial products provided are often relatively standardized. There is a failure to accurately grasp key information such as customers' investment preferences and risk tolerance. There is a lack of services customized according to customers' personalized needs, and there is a problem of not being able to accurately generate sales strategies. Summary of the invention

[0003] The embodiments of the present application provide a method and device for generating a sales strategy, a program product, and a storage medium, so as to at least solve the problem in the related art that a sales strategy cannot be accurately generated.

[0004] According to an embodiment of the present application, a method for generating a sales strategy is provided, comprising: obtaining object data and a target transaction factor of a target object, wherein the object data includes basic object data of the target object and operation data of the target object for products in a target system, and the target transaction factor is determined by performing a correlation analysis on historical data of the products in the target system; inputting the object data and the target transaction factor into a target clustering model to obtain a clustering feature table of the target object, wherein the clustering feature table includes category features and category levels of the target object, and the target clustering model is a model obtained by performing unsupervised training on an initial clustering model using a set of sample object data and a set of sample transaction factors; inputting the clustering feature table into a target regression model to obtain a target feature portrait, wherein the target feature portrait is used to represent the feature distribution of the target object, and the target regression model is a model determined by performing regression analysis modeling on the sample clustering feature table; generating a target sales strategy based on the target feature portrait, wherein the target sales strategy is used to instruct the sale of the products in the target system to the target object.

[0005] In an exemplary embodiment, the unsupervised training process includes: inputting the target sample object data and the target sample transaction factor into the initial clustering model, and the initial clustering model outputs the probability of whether the target sample object belongs to multiple sample clustering features in the target sample clustering feature table, wherein the target sample object data is included in the set of sample object data, and the target sample transaction factor is included in the set of sample transaction factors; and the initial clustering model is trained according to the probability.

[0006] In an exemplary embodiment, the target sample object data and the target sample transaction factor are input into the above-mentioned initial clustering model, and the above-mentioned initial clustering model outputs the probability of whether the target sample object belongs to multiple sample clustering features in the target sample clustering feature table, including: obtaining the above-mentioned set of sample object data and the above-mentioned set of sample transaction factors, wherein the above-mentioned set of sample object data and the above-mentioned set of sample transaction factors are one-to-one corresponding; performing feature processing operations on the above-mentioned set of sample object data and the above-mentioned set of sample transaction factors respectively to obtain a set of sample object features and a set of sample transaction factor features, wherein the above-mentioned feature processing operations include at least one of the following: feature variable processing, extreme value processing, normalization processing, screening processing, and the above-mentioned set of sample object features and the above-mentioned set of sample transaction factor features are both numerically represented; performing feature processing on the above-mentioned set of sample object features and the above-mentioned set of sample transaction factor features The features are respectively subjected to orthogonal transformation operations to obtain a group of first sample object features and a group of first sample transaction factor features, wherein the above-mentioned group of first sample object features and the above-mentioned group of first sample transaction factor features are all linearly independent feature variables; the above-mentioned group of first sample object features and the above-mentioned group of first sample transaction factor features are both subjected to screening operations to obtain a group of screened first sample object features and a group of screened first sample transaction factor features, wherein the cumulative explained variances of the above-mentioned group of screened first sample object features and the above-mentioned group of screened first sample transaction factor features are both greater than a first threshold; the above-mentioned group of screened first sample object features and the above-mentioned group of screened first sample transaction factor features are subjected to iterative clustering operations to output the probability of whether the above-mentioned target sample object belongs to multiple sample clustering features in the target sample clustering feature table to obtain a group of probabilities.

[0007] In an exemplary embodiment, after the initial clustering model is trained according to the probability, the method includes: performing a rating operation on the multiple sample clustering features in the target sample clustering feature table output by the initial clustering model to obtain multiple rating results, wherein the rating results include whether the sample category level corresponding to the sample clustering feature meets the first sample category level; and adjusting the clustering parameters in the initial clustering model based on the multiple rating results.

[0008] In an exemplary embodiment, the regression analysis modeling process includes: extracting multiple sample category features and multiple sample category levels corresponding to multiple sample clustering categories from the sample cluster category feature table; converting the multiple sample category features and the multiple sample category levels into numerical data respectively to obtain multiple first numerical values ​​corresponding to the multiple sample category features and multiple second numerical values ​​corresponding to the multiple sample category levels; iteratively training the initial regression model using the multiple first numerical values ​​and the multiple second numerical values ​​until the initial regression model meets the preset end condition, and then terminating the training to obtain the target regression model.

[0009] In an exemplary embodiment, a target sales strategy is generated based on the target feature portrait, including: determining a target product type corresponding to the target object based on the target feature portrait, wherein the target product type is used to correspond to the risk tolerance and investment propensity of the target object; determining a target product based on the target product type to generate the target sales strategy corresponding to the target product.

[0010] According to an embodiment of the present application, a sales strategy generation device is provided, comprising: a first acquisition module, used to acquire object data and target transaction factors of a target object, wherein the object data includes basic object data of the target object and operation data of the target object for products in a target system, and the target transaction factors are determined by performing correlation analysis on historical data of products in the target system; a first input module, used to input the object data and the target transaction factors into a target clustering model to obtain a clustering feature table of the target object, wherein the clustering feature table includes category features and category levels of the target object, and the target clustering model is a model obtained by performing unsupervised training on an initial clustering model using a set of sample object data and a set of sample transaction factors; a second input module, used to input the clustering feature table into a target regression model to obtain a target feature portrait, wherein the target feature portrait is used to represent the feature distribution of the target object, and the target regression model is a model determined by performing regression analysis modeling on the sample clustering feature table; a first generation module, used to generate a target sales strategy based on the target feature portrait, wherein the target sales strategy is used to instruct the sale of products in the target system to the target object.

[0011] In an exemplary embodiment, the above-mentioned unsupervised training process is trained using a model building device, which includes: a third input module, used to input the target sample object data and the target sample transaction factor into the above-mentioned initial clustering model, and the above-mentioned initial clustering model outputs the probability of whether the target sample object belongs to multiple sample clustering features in the target sample clustering feature table, wherein the above-mentioned set of sample object data includes the above-mentioned target sample object data, and the above-mentioned set of sample transaction factors includes the above-mentioned target sample transaction factor; a first training module, used to train the above-mentioned initial clustering model according to the above-mentioned probability.

[0012] In an exemplary embodiment, the third input module includes: a first acquisition submodule, used to acquire the set of sample object data and the set of sample transaction factors, wherein the set of sample object data and the set of sample transaction factors correspond to each other one by one; a first processing submodule, used to perform feature processing operations on the set of sample object data and the set of sample transaction factors, respectively, to obtain a set of sample object features and a set of sample transaction factor features, wherein the feature processing operations include at least one of the following: feature variable processing, extreme value processing, normalization processing, screening processing, and the set of sample object features and the set of sample transaction factor features are both numerically represented; a first execution submodule, used to perform orthogonal transformation operations on the set of sample object features and the set of sample transaction factor features, respectively, to obtain a set of first sample object features and a set of The first sample transaction factor characteristics, wherein the above-mentioned set of first sample object characteristics and the above-mentioned set of first sample transaction factor characteristics are all linearly independent feature variables; the first screening submodule is used to perform screening operations on the above-mentioned set of first sample object characteristics and the above-mentioned set of first sample transaction factor characteristics to obtain a screened set of first sample object characteristics and a screened set of first sample transaction factor characteristics, wherein the cumulative explained variances of the above-mentioned screened set of first sample object characteristics and the above-mentioned screened set of first sample transaction factor characteristics are both greater than a first threshold; the second execution submodule is used to perform an iterative clustering operation on the above-mentioned screened set of first sample object characteristics and the above-mentioned screened set of first sample transaction factor characteristics, output the probability of whether the above-mentioned target sample object belongs to multiple sample clustering characteristics in the target sample clustering feature table, and obtain a set of probabilities.

[0013] In an exemplary embodiment, the above-mentioned model building device also includes: a first execution module, which is used to perform a rating operation on the above-mentioned multiple sample clustering features in the above-mentioned target sample clustering feature table output by the above-mentioned initial clustering model after training the above-mentioned initial clustering model according to the above-mentioned probability, and obtain multiple rating results, wherein the above-mentioned rating results include whether the sample category level corresponding to the above-mentioned sample clustering feature meets the first sample category level; a first adjustment module, which is used to adjust the clustering parameters in the above-mentioned initial clustering model based on the above-mentioned multiple rating results.

[0014] In an exemplary embodiment, the above-mentioned regression analysis modeling process is established using a model building device, which includes: a first extraction module, used to extract multiple sample category features and multiple sample category levels corresponding to multiple sample clustering categories from the above-mentioned sample clustering category feature table; a first conversion module, used to convert the above-mentioned multiple sample category features and the above-mentioned multiple sample category levels into numerical data, respectively, to obtain multiple first numerical values ​​corresponding to the above-mentioned multiple sample category features and multiple second numerical values ​​corresponding to the above-mentioned multiple sample category levels; a second training module, used to use the above-mentioned multiple first numerical values ​​and the above-mentioned multiple second numerical values ​​to iteratively train the initial regression model until the above-mentioned initial regression model meets the preset end condition, and then terminate the training to obtain the above-mentioned target regression model.

[0015] In an exemplary embodiment, the above-mentioned first generation module includes: a first determination submodule, used to determine the target product type corresponding to the above-mentioned target object based on the above-mentioned target feature portrait, wherein the above-mentioned target product type is used to correspond to the risk tolerance and investment propensity of the above-mentioned target object; a second determination submodule, used to determine the target product based on the above-mentioned target product type to generate the above-mentioned target sales strategy corresponding to the above-mentioned target product.

[0016] According to another embodiment of the present application, a computer program product is provided, including a computer program, wherein the computer program is configured for a processor to execute the steps of any of the above method embodiments.

[0017] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to enable a processor to execute the steps in any of the above method embodiments.

[0018] According to another embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor is configured to execute the computer program to perform the steps in any one of the method embodiments.

[0019] Through this application, the basic object data of the target object, the operation data of the target object on the product in the target system, and the target transaction factor determined by correlation analysis of the historical data of the product in the target system are first obtained, and then the category characteristics and category level of the target object are obtained using the target clustering model, and finally the target feature portrait is obtained based on the category characteristics and category level of the target object using the target regression model, and the target sales strategy is generated using the target feature portrait. Since the generated target sales strategy combines the operation data of the target object on the product and the target transaction factor determined by correlation analysis of the historical data of the product in the target system, the problem of not being able to accurately generate sales strategies in related technologies is solved, and the effect of improving the pertinence of sales strategies is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a hardware structure block diagram of a mobile terminal of a method for generating a sales strategy in an embodiment of the present application;

[0021] Figure 2 is a flow chart of a method for generating a sales strategy according to an embodiment of the present application;

[0022] Figure 3 It is a PCA dimensionality reduction curve diagram of a sales strategy generation method in a specific embodiment of the present application;

[0023] Figure 4 It is a flow chart of a method for generating a sales strategy according to a specific embodiment of the present application;

[0024] Figure 5 It is a data processing flow chart of a method for generating a sales strategy according to a specific embodiment of the present application;

[0025] Figure 6 This is a process of generating a sales strategy in a specific embodiment of the present application. Figure 1 ;

[0026] Figure 7 It is a structural block diagram of a sales strategy generation device according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0029] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal of a method for determining a distribution score in an embodiment of the present application. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0030] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as a computer program corresponding to a method for generating a sales strategy in an embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0031] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0032] In this embodiment, a method for generating a sales strategy is provided. Figure 2 is a flow chart of a method for generating a sales strategy according to an embodiment of the present application, such as Figure 2 As shown, the process includes the following steps:

[0033] Step S202, obtaining object data and a target transaction factor of a target object, wherein the object data includes basic object data of the target object and operation data of the target object for a product in a target system, and the target transaction factor is determined by performing a correlation analysis on historical data of the product in the target system;

[0034] Optionally, the target objects include but are not limited to individual customers and corporate customers with deposit, wealth management, and fund transaction records, such as agency wealth management product customers, self-operated wealth management product customers, in-bank fund customers, and out-bank fund customers.

[0035] Optionally, basic object data includes, but is not limited to, the target object's age, education, occupation, income, assets under management (AUM stratification), AUM allocation (wealth management ratio, fund ratio), and AUM changes. Operational data includes, but is not limited to, the profit and loss situation when buying, holding, selling, and trading, the market changes in the week before and after buying, holding, selling, and trading, product repurchase, holding period, cross-market trading system (CTS) changes, browsing records and consultation records of the main products, and operation records of purchasing key products such as deposits, wealth management, and funds.

[0036] Optionally, the target transaction factors include, but are not limited to, the correlation between transaction changes among major categories of target financial products, the correlation between market trends and the number of target financial product transactions, the correlation between the number of stocks that rise and fall in the market and the number of target financial product transactions within the bank, the correlation between the number of stocks that rise and fall in the market and the number of target financial product transactions outside the bank, and the correlation between market index derivative indicators and the transaction behavior of target objects. Historical data include, but are not limited to, market interest rates and stock market trends, the number of target financial products, market indices, and market derivative indices.

[0037] Step S204, inputting the object data and the target transaction factor into the target clustering model to obtain a clustering feature table of the target object, wherein the clustering feature table includes the category features and category levels of the target object, and the target clustering model is a model obtained by performing unsupervised training on the initial clustering model using a set of sample object data and a set of sample transaction factors;

[0038] Optionally, the category features include, but are not limited to, the target object's consumption frequency, mobile banking account transaction frequency, wage payment amount, fund historical gains / losses, financial product amount, and age. The category level includes, but is not limited to, high-value target object, good-value target object, medium-value target object, low-value target object, conservative, steady, and aggressive.

[0039] Optionally, algorithms for unsupervised training include but are not limited to K-means clustering (K-means for short), hierarchical clustering, and density-based clustering algorithm (Density-Based Spatial Clustering of Applications with Noise for short DBSCAN).

[0040] Step S206, inputting the cluster feature table into the target regression model to obtain a target feature portrait, wherein the target feature portrait is used to represent the feature distribution of the target object, and the target regression model is a model determined by performing regression analysis modeling on the sample cluster feature table;

[0041] Optionally, the target object feature distribution in the target feature portrait can reflect the interaction of several factors such as customer investment preferences, asset distribution, market indicators, etc.

[0042] Step S208, generating a target sales strategy based on the target feature portrait, wherein the target sales strategy is used to instruct the sale of products in the target system to the target object.

[0043] In this embodiment, the execution subject of the above steps may be a terminal, a server, a specific processor set in the terminal or server, or a processor or processing device set relatively independently from the terminal or server, but is not limited thereto. For example, a bank's sales system.

[0044] Through the above steps, the basic object data of the target object, the operation data of the target object on the product in the target system, and the target transaction factor determined by correlation analysis of the historical data of the product in the target system are first obtained, and then the category characteristics and category level of the target object are obtained by using the target clustering model, and finally the target feature portrait is obtained based on the category characteristics and category level of the target object using the target regression model, and the target sales strategy is generated using the target feature portrait. Since the generated target sales strategy combines the operation data of the target object on the product and the target transaction factor determined by correlation analysis of the historical data of the product in the target system, the problem of the inability to accurately generate sales strategies in related technologies is solved, and the effect of improving the pertinence of sales strategies is achieved.

[0045] In an exemplary embodiment, the unsupervised training process includes: inputting the target sample object data and the target sample transaction factor into the initial clustering model, and the initial clustering model outputs the probability of whether the target sample object belongs to multiple sample clustering features in the target sample clustering feature table, wherein the target sample object data is included in the set of sample object data, and the target sample transaction factor is included in the set of sample transaction factors; and the initial clustering model is trained according to the probability.

[0046] Optionally, in this embodiment, during the unsupervised training process, each target sample object may be assigned to multiple sample clustering features, so there is a situation of assignment error, so it is necessary to determine that the target sample object belongs to multiple sample clustering features in the target sample clustering feature table. For example, if a target sample object has a high probability of belonging to a "high-value customer", this means that the probability that the category feature obtained by clustering the target sample object data and the target sample transaction factor belongs to the feature in the target sample clustering feature table and the category feature of "high-value customers" is very high. By considering the probability that the target sample object belongs to multiple sample clustering features, this embodiment can avoid overfitting of specific samples, thereby achieving the purpose of improving the generalization ability of the model.

[0047] In an exemplary embodiment, the target sample object data and the target sample transaction factor are input into the above-mentioned initial clustering model, and the above-mentioned initial clustering model outputs the probability of whether the target sample object belongs to multiple sample clustering features in the target sample clustering feature table, including: obtaining the above-mentioned set of sample object data and the above-mentioned set of sample transaction factors, wherein the above-mentioned set of sample object data and the above-mentioned set of sample transaction factors are one-to-one corresponding; performing feature processing operations on the above-mentioned set of sample object data and the above-mentioned set of sample transaction factors respectively to obtain a set of sample object features and a set of sample transaction factor features, wherein the above-mentioned feature processing operations include at least one of the following: feature variable processing, extreme value processing, normalization processing, screening processing, and the above-mentioned set of sample object features and the above-mentioned set of sample transaction factor features are both numerically represented; performing feature processing on the above-mentioned set of sample object features and the above-mentioned set of sample transaction factor features The features are respectively subjected to orthogonal transformation operations to obtain a group of first sample object features and a group of first sample transaction factor features, wherein the above-mentioned group of first sample object features and the above-mentioned group of first sample transaction factor features are all linearly independent feature variables; the above-mentioned group of first sample object features and the above-mentioned group of first sample transaction factor features are both subjected to screening operations to obtain a group of screened first sample object features and a group of screened first sample transaction factor features, wherein the cumulative explained variances of the above-mentioned group of screened first sample object features and the above-mentioned group of screened first sample transaction factor features are both greater than a first threshold; the above-mentioned group of screened first sample object features and the above-mentioned group of screened first sample transaction factor features are subjected to iterative clustering operations to output the probability of whether the above-mentioned target sample object belongs to multiple sample clustering features in the target sample clustering feature table to obtain a group of probabilities.

[0048] Optionally, in this embodiment, sample object data and sample transaction factors can be obtained from a wide table of sample object information in a sample object database based on a unique identification code of the target sample object, wherein the wide table of sample object information in the sample object database can be obtained by utilizing the bank's production system and big data intelligent platform, purchased external data, and related behavioral data to conduct data exploration, collection, and cleaning.

[0049] Optionally, if the sample objects and sample transaction factors have data in non-numerical format, one-hot encoding can be used to convert them into data in numerical format. By performing feature processing operations on the sample objects and sample transaction factors, the processing process of the target clustering model training and the accuracy of the output can be improved.

[0050] Optionally, the dimensionality reduction operation is implemented by using the principal component analysis algorithm (PCA for short) to illustrate: the sample object features and sample transaction factor features that may be related are converted into a set of linearly unrelated first sample object features and first sample transaction factor features through orthogonal transformation. The first sample object features and the first sample transaction factor features are standardized to ensure that the scale of each feature is consistent; the variance of each feature and the variance contribution rate of each feature are calculated, wherein the variance contribution rate refers to the ratio of the variance of the feature to the sum of the variances of all features, which reflects the degree of explanation of the feature to the total variance of all features; starting from the first feature, the cumulative explained variance of the variance contribution rate of each feature is calculated, and according to the accumulated explained variance, the minimum number of features is found so that the cumulative explained variance reaches or exceeds the first threshold. For example, we reduce the dimension of features such as AUM, number of consumption, number of mobile banking account transactions, amount of wage payment, age, historical fund returns, average monthly balance of financial products, and customer age, and use 95% of the cumulative explained variance as the first threshold. Figure 3 The PCA dimension reduction curve of a sales strategy generation method shown in the figure selects the dimension after dimension reduction to be 7 (horizontal axis), and then uses the PCA method to reduce the dimension (from 9 dimensions to 7 dimensions), and obtains the final first sample object characteristics and first sample transaction factor characteristics: consumption frequency, mobile banking account transaction frequency, salary payment amount, fund historical income / loss, financial product balance, age, and number of views on the main product. This embodiment maximizes the retention of the variance information of the sample object characteristics and the sample transaction factor characteristics in the process of dimension reduction, while reducing the complexity of the sample object characteristics and the sample transaction factor characteristics, thereby achieving the purpose of improving the accuracy of subsequent clustering.

[0051] Optionally, the clustering operation is explained using the K-Means algorithm: Assuming that the 7-dimensional first sample object features and the first sample transaction factor features are obtained after the above-mentioned dimensionality reduction operation, firstly, different categories are formed according to the 7-dimensional first sample object features and the first sample transaction factor features; consumption frequency, mobile banking account transaction frequency, wage payment amount, fund historical income / loss, financial product balance, age, and number of views on the main product. Try clustering from 3 to 11 categories, and record the results of each clustering; then select one or more indicators to evaluate the clustering effect, such as the Silhouette Score, and calculate the silhouette coefficient for each clustering result; finally, analyze the silhouette coefficient to determine which category has the best clustering effect, and then determine the probability that the target sample object belongs to multiple sample clustering features in the target sample clustering feature table, and which category in the clustering result it belongs to. For example, suppose that the bank uses PCA to reduce the dimension, and then uses K-means clustering analysis. The following results may be obtained: Category 1: high-value customers with high AUM, high average monthly balance of financial products, and high wage payment amount. Category 2: Active transaction customers, with high consumption times and mobile banking account transactions, but relatively low AUM. Category 3: Young investors, younger, with medium historical fund returns and average monthly balance of wealth management products. Category 4: Stable customers, with high customer levels, but stable transaction frequency and AUM growth. The probability that target sample object A belongs to high AUM, high average monthly balance of wealth management products, and high amount of salary payment in the target sample clustering feature table is 70%, the probability that the consumption times and mobile banking account transactions are high, but the AUM is relatively low is 45%, the probability that the age is younger, the fund historical returns and the average monthly balance of wealth management products are medium is 20%, and the probability that the customer level is high, but the transaction frequency and AUM growth are stable is 50%. In this embodiment, clustering is used to explore the clustering effects of different categories, identify different target sample object groups, and achieve the purpose of providing personalized services and product recommendations based on their characteristics.

[0052] In an exemplary embodiment, after the initial clustering model is trained according to the probability, the method includes: performing a rating operation on the multiple sample clustering features in the target sample clustering feature table output by the initial clustering model to obtain multiple rating results, wherein the rating results include whether the sample category level corresponding to the sample clustering feature meets the first sample category level; and adjusting the clustering parameters in the initial clustering model based on the multiple rating results.

[0053] Optionally, in this embodiment, the implementation method of the rating operation includes but is not limited to assigning weighted scoring, silhouette coefficient, and label scoring. Assume that sample category level category 1 corresponds to the first sample category level A high-value target sample object, sample category level category 2 corresponds to B good-value target sample object, sample category level category 3 corresponds to the first sample category level C medium-value target sample object, and sample category level category corresponds to the first sample category level D low-value target sample object. For example, the probability that the target sample object A belongs to the high AUM, high average monthly balance of financial products, and high amount of salary paid on behalf of the target sample clustering feature table is 70%, and the rating obtained after performing the rating operation is high, that is, the target sample object A belongs to the sample category level category 1, which meets the first sample category level A high-value target object. If the category level of some target sample objects does not meet expectations (for example, some high-value customers are mistakenly classified as "low-value customers"), the parameters of the clustering model, such as the number of features, the feature weight of each feature, etc., can be adjusted according to these rating results. This embodiment achieves the purpose of making the clustering results more in line with actual business needs by ensuring that the category level of the sample clustering feature meets the predefined standards. In banking, this iteratively optimized clustering model can more accurately identify the characteristics of different categories of customers, thereby providing customized financial products and services.

[0054] In an exemplary embodiment, the regression analysis modeling process includes: extracting multiple sample category features and multiple sample category levels corresponding to multiple sample clustering categories from the sample cluster category feature table; converting the multiple sample category features and the multiple sample category levels into numerical data respectively to obtain multiple first numerical values ​​corresponding to the multiple sample category features and multiple second numerical values ​​corresponding to the multiple sample category levels; iteratively training the initial regression model using the multiple first numerical values ​​and the multiple second numerical values ​​until the initial regression model meets the preset end condition, and then terminating the training to obtain the target regression model.

[0055] Optionally, in this embodiment, a single-hot encoding can be used to convert multiple sample category features and multiple sample category levels into numerical data, and the preset end conditions include but are not limited to the accuracy of the first and second numerical values ​​in predicting the behavior of the target sample object, AUC, and the number of iterations. For example, high consumption frequency corresponds to 0, low consumption frequency corresponds to 1, high fund historical returns correspond to 2, high fund historical losses correspond to 3, low page views of the main product correspond to 6, and high page views of the main product correspond to 5. Sample category level category 1 corresponds to value 1, and sample category level category 2 corresponds to value 2. Logistic regression or other suitable regression models are used as the initial model, and high consumption frequency corresponds to 0, high fund historical returns correspond to 2, and high page views of the main product correspond to 5. Sample category level category 1 high-value target objects are matched to value 1, and the target sample object behavior is predicted to be purchase, which is consistent with the actual behavior; low consumption frequency corresponds to 1, high fund historical losses correspond to 3, and high page views of the main product correspond to 5, which is a sample category level category 2 active trading customer, and the target sample object behavior is predicted to be purchase, which is consistent with the actual behavior, and the accuracy rate is 100%, and the final regression model is obtained.

[0056] In an exemplary embodiment, a target sales strategy is generated based on the target feature portrait, including: determining a target product type corresponding to the target object based on the target feature portrait, wherein the target product type is used to correspond to the risk tolerance and investment propensity of the target object; determining a target product based on the target product type to generate the target sales strategy corresponding to the target product.

[0057] Optionally, in this embodiment, assuming that the target feature profile has high consumption frequency, high fund historical returns, and high views of the main product, and the target sample object behavior is purchase, then the target product type of the active target object is high risk and high return, and the target sales strategy of holding online seminars can be used to recommend stocks and stocks or hedge funds with high risk and high return. This embodiment determines the target product type corresponding to the target object based on the target feature profile, and recommends products to the target object based on the target product type, thereby achieving the effect of improving the pertinence of the sales strategy.

[0058] The present invention will be described below in conjunction with specific embodiments:

[0059] This specific embodiment is described by taking the sale of products to individual customer A as an example. Figure 4 This is a flow chart of a method for generating a sales strategy according to a specific embodiment of the present application, comprising the following steps:

[0060] S402, using the bank's production system and big data intelligent platform, obtain the internal data and external data of multiple individual customers (corresponding to the above-mentioned target sample object data and target sample transaction factors), conduct data exploration, collection, cleaning and processing, and establish a wide table of financial product transaction customer information with the individual customer number as the unique identification code (excluding sensitive information such as customer name, mobile phone number, ID number, address, and excluding minors' data). After the wide table is established, it is stored in the laboratory basic database. Among them, individual customers include individual customers with deposit, wealth management, and fund transaction records, agency sales of wealth management products, self-operated wealth management products, in-bank fund customers, and out-of-bank fund customers. Internal data includes but is not limited to the individual customer's age, education, occupation, income, assets under management (Assets Under Management, referred to as AUM stratification), AUM allocation (wealth management ratio, fund ratio), AUM changes, profit and loss at the time of buying, holding, selling, and transaction, market changes in the week before and after buying, holding, selling, and transaction, product repurchase, holding period, cross-market trading system (Cross-Trading External data include, but are not limited to, the correlation between changes in transactions between major categories of financial products for individual customers, the correlation between market trends and the number of target objects for financial product transactions, the correlation between the number of stocks that rise and fall in the market and the number of target objects for financial product transactions within the bank, the correlation between the number of stocks that rise and fall in the market and the number of target objects for financial product transactions outside the bank, the correlation between market index derivative indicators and the transaction behavior of target objects, market interest rates and stock market trends, the number of target objects for different financial products, market indices, and market derivative indices;

[0061] S404, perform variable processing, extreme value processing, normalization and other feature processing operations on internal data and external data, and determine the input variables according to correlation analysis and dimensionality reduction operations, such as Figure 5 As shown, Figure 5 This is a data processing flow chart of a method for generating a sales strategy according to a specific embodiment of the present application:

[0062] S502, examine the internal and external data, check the variable data distribution, maximum value, minimum value, mean value, median value, and check whether some data are unreasonable (such as age 130 years old); fill in the blanks and delete unreasonable variables;

[0063] S504, the data is divided into continuous variables and categorical variables. The categorical variables are converted into model-entry variables through hot coding to reduce information loss. The continuous variables are normalized according to the nature of the variables, the number of transactions, and the frequency type. The assets and liabilities are processed by extreme value and divided into 5-9 grades according to the difference in data intervals to obtain discrete variables and categorical variables.

[0064] S506, analyze the correlation of variables through Pearson and other methods, then reduce the dimension of the variables, and use the target clustering model to implement the dimension reduction operation with the principal component analysis algorithm to illustrate: convert the possibly related internal data and external data into a set of linearly unrelated internal data and external data through orthogonal transformation. Standardize the internal and external data to ensure that the scale of each internal and external data is consistent; calculate the variance of each internal and external data and the variance contribution rate of each internal and external data, calculate the cumulative explained variance of the variance contribution rate of each data, and find the minimum number of data based on the accumulated explained variance so that the cumulative explained variance reaches or exceeds the first threshold. For example, reduce the dimension of features such as AUM, number of consumption times, number of mobile banking account transactions, amount of wage payment, age, historical fund returns, average monthly balance of wealth management products, and customer age, and use 95% of the cumulative explained variance as the first threshold. Figure 3 The PCA dimension reduction curve of a sales strategy generation method shown in the figure selects the dimension after dimension reduction to be 7 (horizontal axis), and then uses the PCA method to reduce the dimension (from 9 dimensions to 7 dimensions);

[0065] S508, obtaining the final model input variables: consumption frequency, mobile banking account transaction frequency, salary payment amount, fund historical gains / losses, financial product balance, age, and number of views on the main product;

[0066] S406, to profile the customer base, the main methods used are: 1) using ready-made label fields for statistics and analyzing the distribution of each label of the customer; 2) using the target clustering model and target regression model, clustering the customers based on the customers’ numerical input variables, and then performing statistical analysis on the customers divided by the model, using statistical features to describe each cluster category.

[0067] In this embodiment, the target clustering model is used to perform clustering operations using the K-Means algorithm: different clustering categories are tried for clustering, and then the silhouette coefficient is used to evaluate the clustering effect, and the number of categories for the final clustering is determined based on this experiment. Assume that the point with the highest silhouette coefficient is when the number of categories is 5, so 5 is selected as the number of clusters, as shown in Table 1.

[0068] Table 1:

[0069]

[0070] Considering the differences between entities, we use the analytic hierarchy process to weight the variables in the model, based on the consumption frequency, mobile banking account transaction frequency, salary payment amount, fund historical gains / losses, financial product balance, age, and the number of views on the main products. We then rate individual customers in each dimension in a segmented scoring manner, and sum up the scoring results of each dimension according to a certain weight, and perform data standardization to obtain the individual customer rating, as shown in Table 2.

[0071] Table 2:

[0072]

[0073] The target regression model is used to conduct statistical analysis on the individual customer categories divided by the target clustering model, and statistical characteristics are used to describe each clustering category. For example, the next stage of financial transaction behavior of individual customers of categories 0 and 2 may be purchasing, the next stage of financial transaction behavior of individual customers of category 1 may be holding, the next stage of financial transaction behavior of individual customers of category 3 may be selling, and the next stage of financial transaction behavior of individual customers of category 4 may be no reaction.

[0074] S408, the customer sales model should be able to reflect the interaction of several factors such as individual customer investment preferences, asset distribution, market indicators, etc., and can combine the predicted financial transaction behavior, the characteristics of the products already held and the recommended products to generate targeted sales strategies, such as Figure 6 As shown, Figure 6 This is a process of generating a sales strategy in a specific embodiment of the present application. Figure 1 :

[0075] S602: Divide the model input variables into training set (67%), test set (25%), and validation set (8%) using PYTHON program; classify individual customers into: regular customers with financial product transaction records in the past year, ordinary customers who hold financial products or have transaction records in the past three years, and unregistered customers with no holding records or transaction records;

[0076] S604: Model training: For familiar customers, they have rich traffic activities, financial products and transaction data: for some scenarios, they can use logistic regression models, XGB models, and decision trees, which can fully refer to a series of characteristic data such as customer age, gender, liabilities, financial products held, and transaction behavior, supplement effective information, and make the prediction more accurate, especially in batch sales. The success rate is also greatly improved. For ordinary customers, they use logistic regression models, XGB models, and collaborative filtering for training and optimization. For unregistered customers, based on the basic assumption that "customers in the same customer group have high similarity in investment preferences", analyze the comprehensive situation of the customer's customer group's investment and behavior preferences, combine the personal characteristics of individual customers, and use collaborative filtering to obtain the customer's investment products and behavior preferences, and formulate corresponding intelligent matching strategies. The result of the model output is the customer's next stage sales strategy; through optimization, the model with better KS and AUC indicators is finally selected, and the model is stored in the model library to prepare for model training, and the F prediction and matching results are pushed to the sales platform or customer manager workbench in real time; model training, use the validation set for verification, and select the optimal model corresponding to each category of individual customers.

[0077] S606: Model packaging and application: Model packaging: Combine the variable feature library and the optimal model, and use the Python function to package them into a .py file in the my_package directory or use I MPORT to package them into a pickle file, in preparation for the sales staff to apply the model in the next stage. Use the predicted input variables to identify the customer type, and select the corresponding optimal model to match the customer's financial transaction behavior in the next stage. Comprehensively consider customer classification, risk preference, product preference, etc., and output sales strategies.

[0078] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0079] In the present embodiment, a sales strategy generation device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0080] Figure 7 is a structural block diagram of a sales strategy generation device according to an embodiment of the present application, such as Figure 7 As shown, the device comprises:

[0081] A first acquisition module 702 is used to acquire object data and a target transaction factor of a target object, wherein the object data includes basic object data of the target object and operation data of the target object on a product in a target system, and the target transaction factor is determined by performing a correlation analysis on historical data of the product in the target system;

[0082] A first input module 704 is used to input the object data and the target transaction factor into a target clustering model to obtain a clustering feature table of the target object, wherein the clustering feature table includes the category features and category levels of the target object, and the target clustering model is a model obtained by performing unsupervised training on an initial clustering model using a set of sample object data and a set of sample transaction factors;

[0083] The second input module 706 is used to input the clustering feature table into the target regression model to obtain a target feature portrait, wherein the target feature portrait is used to represent the feature distribution of the target object, and the target regression model is a model determined by performing regression analysis modeling on the sample clustering feature table;

[0084] The first generating module 708 is used to generate a target sales strategy based on the target feature portrait, wherein the target sales strategy is used to instruct the sale of the product in the target system to the target object.

[0085] In an exemplary embodiment, the above-mentioned unsupervised training process is trained using a model building device, which includes: a third input module, used to input the target sample object data and the target sample transaction factor into the above-mentioned initial clustering model, and the above-mentioned initial clustering model outputs the probability of whether the target sample object belongs to multiple sample clustering features in the target sample clustering feature table, wherein the above-mentioned set of sample object data includes the above-mentioned target sample object data, and the above-mentioned set of sample transaction factors includes the above-mentioned target sample transaction factor; a first training module, used to train the above-mentioned initial clustering model according to the above-mentioned probability.

[0086] In an exemplary embodiment, the third input module includes: a first acquisition submodule, used to acquire the set of sample object data and the set of sample transaction factors, wherein the set of sample object data and the set of sample transaction factors correspond to each other one by one; a first processing submodule, used to perform feature processing operations on the set of sample object data and the set of sample transaction factors, respectively, to obtain a set of sample object features and a set of sample transaction factor features, wherein the feature processing operations include at least one of the following: feature variable processing, extreme value processing, normalization processing, screening processing, and the set of sample object features and the set of sample transaction factor features are both numerically represented; a first execution submodule, used to perform orthogonal transformation operations on the set of sample object features and the set of sample transaction factor features, respectively, to obtain a set of first sample object features and a set of The first sample transaction factor characteristics, wherein the above-mentioned set of first sample object characteristics and the above-mentioned set of first sample transaction factor characteristics are all linearly independent feature variables; the first screening submodule is used to perform screening operations on the above-mentioned set of first sample object characteristics and the above-mentioned set of first sample transaction factor characteristics to obtain a screened set of first sample object characteristics and a screened set of first sample transaction factor characteristics, wherein the cumulative explained variances of the above-mentioned screened set of first sample object characteristics and the above-mentioned screened set of first sample transaction factor characteristics are both greater than a first threshold; the second execution submodule is used to perform an iterative clustering operation on the above-mentioned screened set of first sample object characteristics and the above-mentioned screened set of first sample transaction factor characteristics, output the probability of whether the above-mentioned target sample object belongs to multiple sample clustering characteristics in the target sample clustering feature table, and obtain a set of probabilities.

[0087] In an exemplary embodiment, the above-mentioned model building device also includes: a first execution module, which is used to perform a rating operation on the above-mentioned multiple sample clustering features in the above-mentioned target sample clustering feature table output by the above-mentioned initial clustering model after training the above-mentioned initial clustering model according to the above-mentioned probability, and obtain multiple rating results, wherein the above-mentioned rating results include whether the sample category level corresponding to the above-mentioned sample clustering feature meets the first sample category level; a first adjustment module, which is used to adjust the clustering parameters in the above-mentioned initial clustering model based on the above-mentioned multiple rating results.

[0088] In an exemplary embodiment, the above-mentioned regression analysis modeling process is established using a model building device, which includes: a first extraction module, used to extract multiple sample category features and multiple sample category levels corresponding to multiple sample clustering categories from the above-mentioned sample clustering category feature table; a first conversion module, used to convert the above-mentioned multiple sample category features and the above-mentioned multiple sample category levels into numerical data, respectively, to obtain multiple first numerical values ​​corresponding to the above-mentioned multiple sample category features and multiple second numerical values ​​corresponding to the above-mentioned multiple sample category levels; a second training module, used to use the above-mentioned multiple first numerical values ​​and the above-mentioned multiple second numerical values ​​to iteratively train the initial regression model until the above-mentioned initial regression model meets the preset end condition, and then terminate the training to obtain the above-mentioned target regression model.

[0089] In an exemplary embodiment, the above-mentioned first generation module includes: a first determination submodule, used to determine the target product type corresponding to the above-mentioned target object based on the above-mentioned target feature portrait, wherein the above-mentioned target product type is used to correspond to the risk tolerance and investment propensity of the above-mentioned target object; a second determination submodule, used to determine the target product based on the above-mentioned target product type to generate the above-mentioned target sales strategy corresponding to the above-mentioned target product.

[0090] An embodiment of the present application provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.

[0091] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.

[0092] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.

[0093] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0094] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0095] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.

[0096] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0097] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for generating a sales strategy, characterized in that: include: Acquire object data and a target transaction factor of a target object, wherein the object data includes basic object data of the target object and operation data of the target object for a product in a target system, and the target transaction factor is determined by performing a correlation analysis on historical data of the product in the target system; Inputting the object data and the target transaction factor into a target clustering model to obtain a clustering feature table of the target object, wherein the clustering feature table includes the category features and category levels of the target object, and the target clustering model is a model obtained by performing unsupervised training on an initial clustering model using a set of sample object data and a set of sample transaction factors; Inputting the cluster feature table into the target regression model to obtain a target feature portrait, wherein the target feature portrait is used to represent the feature distribution of the target object, and the target regression model is a model determined by performing regression analysis modeling on the sample cluster feature table; A target sales strategy is generated based on the target feature portrait, wherein the target sales strategy is used to instruct the sale of the product in the target system to the target object.

2. The method according to claim 1, characterized in that The unsupervised training process includes: Inputting target sample object data and target sample transaction factors into the initial clustering model, and outputting the probability of whether the target sample object belongs to a plurality of sample clustering features in the target sample clustering feature table by the initial clustering model, wherein the set of sample object data includes the target sample object data, and the set of sample transaction factors includes the target sample transaction factor; The initial clustering model is trained according to the probability.

3. The method according to claim 2, characterized in that The target sample object data and the target sample transaction factor are input into the initial clustering model, and the initial clustering model outputs the probability of whether the target sample object belongs to a plurality of sample clustering features in the target sample clustering feature table, including: Acquire the set of sample object data and the set of sample transaction factors, wherein there is a one-to-one correspondence between the set of sample object data and the set of sample transaction factors; Performing feature processing operations on the set of sample object data and the set of sample transaction factors respectively to obtain a set of sample object features and a set of sample transaction factor features, wherein the feature processing operation includes at least one of the following: feature variable processing, extreme value processing, normalization processing, and screening processing, and the set of sample object features and the set of sample transaction factor features are both numerically represented; Performing orthogonal transformation operations on the set of sample object features and the set of sample transaction factor features respectively to obtain a set of first sample object features and a set of first sample transaction factor features, wherein the set of first sample object features and the set of first sample transaction factor features are both linearly unrelated feature variables; Performing a screening operation on the set of first sample object features and the set of first sample transaction factor features to obtain a screened set of first sample object features and a screened set of first sample transaction factor features, wherein the cumulative explained variances of the screened set of first sample object features and the screened set of first sample transaction factor features are both greater than a first threshold; An iterative clustering operation is performed on the screened set of first sample object features and the screened set of first sample transaction factor features, and the probability of whether the target sample object belongs to multiple sample clustering features in the target sample clustering feature table is output to obtain a set of probabilities.

4. The method according to claim 2, characterized in that: After training the initial clustering model according to the probability, the method includes: Performing a rating operation on the multiple sample clustering features in the target sample clustering feature table output by the initial clustering model to obtain multiple rating results, wherein the rating results include whether the sample category level corresponding to the sample clustering feature meets the first sample category level; The clustering parameters in the initial clustering model are adjusted based on the multiple rating results.

5. The method according to claim 1, characterized in that The process of regression analysis modeling includes: Extracting a plurality of sample category features and a plurality of sample category levels corresponding to a plurality of sample cluster categories from the sample cluster category feature table; Respectively converting the plurality of sample category features and the plurality of sample category levels into numerical data to obtain a plurality of first numerical values ​​corresponding to the plurality of sample category features and a plurality of second numerical values ​​corresponding to the plurality of sample category levels; The initial regression model is iteratively trained using the multiple first values ​​and the multiple second values ​​until the initial regression model meets a preset end condition, and the training is terminated to obtain the target regression model.

6. The method according to claim 1, characterized in that Generating a target sales strategy based on the target feature portrait includes: Determining a target product type corresponding to the target object based on the target feature portrait, wherein the target product type is used to correspond to the risk tolerance and investment propensity of the target object; A target product is determined based on the target product type to generate the target sales strategy corresponding to the target product.

7. A sales strategy generating device, characterized in that: include: A first acquisition module is used to acquire object data and a target transaction factor of a target object, wherein the object data includes basic object data of the target object and operation data of the target object on a product in a target system, and the target transaction factor is determined by performing a correlation analysis on historical data of the product in the target system; A first input module is used to input the object data and the target transaction factor into a target clustering model to obtain a clustering feature table of the target object, wherein the clustering feature table includes the category features and category levels of the target object, and the target clustering model is a model obtained by performing unsupervised training on an initial clustering model using a set of sample object data and a set of sample transaction factors; A second input module is used to input the cluster feature table into a target regression model to obtain a target feature portrait, wherein the target feature portrait is used to represent the feature distribution of the target object, and the target regression model is a model determined by performing regression analysis modeling on the sample cluster feature table; The first generating module is used to generate a target sales strategy based on the target feature portrait, wherein the target sales strategy is used to instruct the sale of the product in the target system to the target object.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 6 when executed by a processor.

10. An electronic 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 steps of the method described in any one of claims 1 to 6 are implemented.

Images