Security customer analysis method, storage medium and computer equipment

Abstract

The invention discloses a security customer analysis method, a storage medium and computer equipment. The security customer analysis method based on the RFM model comprises the following steps: acquiring asset data and transaction data of security customers; constructing a preference score matrix between the user and the product based on the asset data and the transaction data; constructing a preference propagation network based on the preference score matrix between the user and the product; calculating a target probability based on a preference value output by the preference propagation network and the expanded RFM model; the extended RFM model introduces three factors of transaction popularity, asset score and return score into asset matching parameters. Through the above mode, the interaction behavior analysis between the security user and the product can be accurately carried out.

Description

Technical Field

[0001] The present application relates to the field of data analysis, and in particular to a securities customer analysis method, storage medium and computer equipment. Background Art

[0002] In the securities industry, customer segmentation and value assessment are key steps to achieve precision marketing and service optimization. Although the traditional RFM (Recency, Frequency, Monetary) model can initially classify customers, its limitation is that it fails to fully consider customers' dynamic trading behavior and asset allocation preferences. Therefore, a more comprehensive method is needed to improve the accuracy and practicality of customer segmentation and value assessment. Summary of the invention

[0003] This application mainly provides a securities customer analysis method, storage medium and computer equipment to solve the problem that the RFM model is not accurate enough in analyzing securities customers.

[0004] To solve the above technical problems, a technical solution adopted in the present application is: to provide a securities client analysis method based on the RFM model, including: obtaining asset data and transaction data of securities clients; constructing a preference score matrix between users and products based on the asset data and the transaction data; constructing a preference propagation network based on the preference score matrix between users and products; calculating the target probability based on the preference value output by the preference propagation network and the expanded RFM model; the expanded RFM model introduces three factors of transaction heat, asset score and return score into the asset matching parameters.

[0005] In some embodiments, constructing a preference score matrix between users and products based on the asset data and the transaction data includes: constructing an association matrix between users and products based on the asset data and the transaction data; calculating the preference similarity between users based on the association matrix between users and products; calculating the degree of user preference for products based on the preference similarity between users and the users' actual transaction behaviors, and generating a preference score matrix.

[0006] In some embodiments, the formula for calculating the preference degree is:

[0007] Among them, the preference degree (u i , p k ) represents user u i For product k degree of preference, preference similarity (u i ,u j ) represents user u i With user uj The similarity of preferences between them, transaction amount (u j , p k ) represents user u i For product k The transaction amount.

[0008] In some embodiments, the calculation of the user's preference for a product by combining the preference similarity and the user's actual transaction behavior also includes: obtaining a preset preference weight; the preference weight is set based on business needs, the asset data, and the analysis results of the transaction data; and adjusting the user's preference for the product based on the preference weight to generate a weighted preference score matrix.

[0009] In some embodiments, the construction of a preference propagation network based on the preference score matrix between the user and the product includes: initializing the network structure according to the association matrix between the user and the product and the preference similarity between the users; setting an initial preference value for each user node in the network structure; and defining a rule for preference propagation in the network structure based on the initial preference value to form a preference propagation network.

[0010] In some embodiments, the rules for propagating preferences in the network structure are defined based on the initial preference values ​​to form a preference propagation network, including: updating the user's preference values ​​for products through iterative calculations; and stopping the iterative calculations in response to the preference values ​​satisfying a preset termination condition.

[0011] In some embodiments, the formula for the iterative calculation is:

[0012] Among them, the new preference value (u i ) indicates that it is used to update user u i The new preference value of the initial preference value, α is the relative importance of controlling the initial preference value and the propagated preference, N(u i ) indicates that the i The set of all adjacent user nodes, N (u j ) indicates that the j The set of all adjacent user nodes, ω(u i ,u j ) represents user u i With user u j The edge weight between j ,u k ) represents user u j With user u k The edge weight between ; where α is greater than 0 and less than 1.

[0013] In some embodiments, the target probability is calculated based on the preference value output by the preference propagation network and the expanded RFM model, including: constructing a probability model for each user to select each product according to the preference value of each user node in the preference propagation network to generate a selection probability matrix for each user to select each product; integrating the selection probability matrix and the expanded RFM model parameters to perform hierarchical clustering to assign cluster labels to each user; analyzing the characteristics of each cluster in the hierarchical clustering to identify typical user behavior patterns.

[0014] In order to solve the above technical problems, another technical solution adopted by the present application is: providing a storage medium on which program data is stored, and when the program data is executed by a processor, the steps of the securities client analysis method as described above are implemented.

[0015] The present application also provides a computer device, including a processor and a memory connected to each other, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the securities client analysis method as described above are implemented.

[0016] The beneficial effects of the present application are as follows: Different from the prior art, the present application discloses a securities client analysis method, storage medium and computer equipment. By acquiring the asset data and transaction data of securities clients; constructing a preference score matrix between users and products based on the asset data and transaction data to reflect the user's transaction status of the product; constructing a preference propagation network based on the preference score matrix between users and products to simulate the preference propagation process between users to more accurately predict the user's potential interest in unpurchased products; calculating the target probability based on the preference value output by the preference propagation network and the extended RFM model; the extended RFM model introduces three factors, namely transaction heat, asset score and return score, into the asset matching parameters to convert the user's preference into a specific decision, thereby dynamically adjusting the clustering method according to the visualization results and business needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work, among which: Figure 1 It is a flow chart of an embodiment of a securities client analysis method provided by the present application; Figure 2 Yes Figure 1 The flowchart of the method step 200 is shown as an embodiment; Figure 3 Yes Figure 2 The flowchart of the embodiment of the method step 230 is shown; Figure 4 Yes Figure 1 The flowchart of the method step 300 is shown as an embodiment; Figure 5 Yes Figure 1 The flowchart of the embodiment of the method step 330 is shown; Figure 6 Yes Figure 1 The flowchart of the method step 400 is shown as an embodiment; Figure 7 It is a structural schematic diagram of an embodiment of a storage medium provided by the present application; Figure 8 It is a structural diagram of an embodiment of a computer device provided by the present application. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] The terms "first", "second", "third" in the embodiments of the present application are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first", "second", "third" can expressly or implicitly include at least one of the features. In the description of the present application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes steps or units that are not listed, or optionally also includes other steps or units inherent to these processes, methods, products or devices.

[0020] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0021] See also Figure 1, Figure 1 : is a flow chart of an embodiment of a securities customer analysis method provided by the present application. The securities customer analysis method based on the RFM model includes: 100: Obtain asset data and transaction data of securities customers.

[0022] Collect and organize the asset data of securities clients, including historical asset allocation data, such as the holding ratio of various assets, transaction frequency, etc. Obtain the transaction data of securities users, including transaction behavior records, such as the transaction behavior of customers, such as buying, selling, holding time records, etc.

[0023] Optionally, the acquired asset data and transaction data are cleansed and preprocessed to ensure data integrity and consistency.

[0024] 200: Build a preference score matrix between users and products based on asset data and transaction data.

[0025] A preference score matrix between users and products is constructed based on asset data and transaction data. The preference score matrix records the user's preference scores for products and can reflect the user's transaction tendency for products.

[0026] Further, see Figure 2 , 200 including: 210: Build a correlation matrix between users and products based on asset data and transaction data.

[0027] Aggregate the collected transaction data, summarize the transaction amounts by customer and product, and generate a correlation matrix between users and products, such as the following table:

[0028] In the association matrix between users and products, each row represents a user, such as u1, u2, u3; each column represents a product, such as p1, p2, p3, p4; the value in the matrix represents the total transaction amount of the user for the product, such as user 1 spends 10,000 on product 1, user 2 spends 15,000 on product 3, etc.

[0029] 220:Calculate the preference similarity between users based on the association matrix between users and products.

[0030] Specifically, the preference similarity between users is calculated by the following formula:

[0031] Among them, preference similarity (u i ,u j) represents the preference similarity between user ui and user uj; Ai and Aj represent the preference vectors of user ui and user uj respectively, that is, the vector composed of the transaction amounts of user ui and user uj on all products.

[0032] Among them, Ai·Aj can be calculated by the following formula:

[0033] Among them, A i，p and A j，p Represents user u i and user u j The transaction amount for product p; A i ·A j For each product p, user u i and user u j Multiply the transaction amounts on that product and add the results for all products.

[0034] Among them, ||Ai|| and ||Aj|| represent the Euclidean norm of vector Ai and vector Aj respectively, and the calculation method is as follows: and .

[0035] Vector Ai and vector Aj represent the preference vector for each user. The transaction amount for each product is squared, summed, and then the square root is taken to obtain the Euclidean norm of the vector.

[0036] The similarity score between user ui and user uj is obtained through the above calculation. The similarity score ranges from [-1, 1], where 1 means they are exactly the same and 0 means there is no correlation.

[0037] For example, there are two users u1 and u3, and four products p1, p2, p3, p4. The transaction amounts of users u1 and u3 on these four products are as follows: User u1: [A 1，1 , A 1，2 , A 1，3 , A 1，4 ]=[10000, 5000, 0, 0], User u3: [A 3，1 , A 3，2 , A 3，3 , A 3，4 ]=[20000, 0, 0, 0].

[0038] To calculate the preference similarity between users u1 and u3, follow these steps: First, calculate the dot product: ; Compute the Euclidean norm of each vector from the dot product: , ; Calculate the preference similarity between u1 and u3: .

[0039] 230: Based on the preference similarity between users and their actual transaction behaviors, calculate the user's preference for the product and generate a preference score matrix.

[0040] Specifically, the formula for calculating the degree of preference is:

[0041] Among them, the preference degree (u i , p k ) represents user u i For product k degree of preference, preference similarity (u i ,u j ) represents user u i With user u j The similarity of preferences between them, transaction amount (u j , p k ) represents user u i For product k The transaction amount.

[0042] The user's preference for the product is calculated by matrix multiplication to generate a preference score matrix. Each row in the preference score matrix represents a user, each column represents a product, and the value in the preference score matrix represents the user's preference for the product.

[0043] Optionally, see Figure 3 , 230 also includes: 231: Get the preset preference weights. The preference weights are set based on the analysis results of business requirements, asset data, and transaction data.

[0044] Assign weights to users' preferences for products to ensure that the importance of different factors is reasonably reflected.

[0045] Assign appropriate weights to each factor based on business needs and data analysis results.

[0046] For example, the weights are set as follows: weight of preference similarity between users: 0.7, weight of actual transaction amount of users: 0.3.

[0047] 232: Adjust the user's preference for the product according to the preference weight and generate a weighted preference score matrix.

[0048] For example, when the weight of preference similarity between users is 0.7 and the weight of actual transaction amount of users is 0.3, the weighted preference of users for products is calculated by the following formula: .

[0049] According to the set weights, the user's preference for the product is adjusted to generate a weighted preference score matrix. Each row of the weighted preference score matrix represents a user, each column represents a product, and the value in the matrix represents the user's weighted preference for the product.

[0050] 300: Construct a preference propagation network based on the preference score matrix between users and products.

[0051] The preference propagation network is used to simulate the preference propagation process among users to more accurately predict users' potential interests in unpurchased products.

[0052] The construction of the preference propagation network depends on the similarity between users and the known interaction information between users and products.

[0053] Further, see Figure 4 , 300 including: 310: Initialize the network structure based on the association matrix between users and products and the preference similarity between users.

[0054] Use the networkx library to create a network where nodes represent users or products and edges represent the strength of relationships between users and products, and between users.

[0055] The network structure is initialized based on the association matrix between users and products and the user similarity matrix. Each node in the network represents a user or a product, and the weight of the edge represents the transaction amount between the user and the product, or the similarity score between users.

[0056] 320: Setting an initial preference value for each user node in the network structure.

[0057] Operate the node attributes of the networkx graph object, traverse the association matrix between users and products, calculate the total transaction amount of each user as its initial preference value, and store it in the network node attribute. Each user node now has an initial preference value, which reflects the user's currently known product preference.

[0058] Specifically, the initial preference value is calculated by the following formula: .

[0059] 330: Based on the initial preference value, a rule for propagating the preference in the network structure is defined to form a preference propagation network.

[0060] Define the rules by which preferences are propagated across the network. These rules determine how preferences are passed from one node to another.

[0061] Further, see Figure 5 , 330 also includes: 331: Update the user's preference value for the product through iterative calculation.

[0062] Specifically, the formula for iterative calculation is:

[0063] Among them, the new preference value (u i ) indicates that it is used to update user u i The new preference value of the initial preference value, α is the relative importance of controlling the initial preference value and the propagated preference, N(u i ) indicates that the i The set of all adjacent user nodes, N (u j ) indicates that the j The set of all adjacent user nodes, ω(u i ,u j ) represents user u i With user u j The edge weight between j ,u k ) represents user u j With user u k The edge weights between .

[0064] Here, α is greater than 0 and less than 1.

[0065] The above formula is used to update the preference value of each user node, taking into account the initial preference and the preference propagation from neighboring nodes. The updated preference value can better reflect the user's interest in products that he has not been exposed to.

[0066] 332: In response to the preference value satisfying a preset termination condition, stop the iterative calculation.

[0067] The user's preference parameters for products are updated through iterative calculations to simulate the propagation of user preferences in the network until the preference values ​​converge to a stable state. The iteration termination condition is set to ensure that the algorithm stops when the preference values ​​converge or the maximum number of iterations is reached.

[0068] Set an initial preference value for each user node and copy it as the old preference value for comparison. Iterate and update the preference value of each user node according to the preference propagation rule until the termination condition is met.

[0069] Furthermore, the preference value of each user node is updated according to the preference propagation formula, and the convergence is checked. If the preference value does not converge, data is prepared for the next round of iteration; otherwise, the iteration is terminated.

[0070] Furthermore, preference propagation updates are performed within the maximum number of iterations until the preference value converges, and finally a stable user preference value is obtained.

[0071] Specifically, the difference between the new preference value and the old preference value is calculated, and the iterative calculation is stopped after the difference satisfies a preset stop condition.

[0072] The formula for judging iterative convergence is:

[0073] Among them, ε is a small positive number, which indicates the maximum allowable error range; if the preference changes of all users are less than ε, the preference values ​​are considered to have converged.

[0074] Traverse each node in the network to determine whether the preference value change of each node in the network meets the preset range; in response to the preference value change of each node in the network meeting the preset range, return a Boolean value indicating that the preference value has converged.

[0075] Or, in response to a preference value change of an existing node in the network not meeting a preset range, a Boolean value is returned, indicating that the preference value has not completed convergence.

[0076] 400: Calculate the target probability based on the preference value output by the preference propagation network and the extended RFM model. The extended RFM model introduces three factors: transaction heat, asset score and return score in the asset matching parameters.

[0077] Based on the preference parameters finally obtained in the preference propagation network, the probability of the user choosing a specific product is calculated. This step is the key link in converting the user's preferences into specific decisions.

[0078] Specifically, the calculation method of the extended parameters includes: Asset score = Ʃ (amount of this type of assets ÷ total asset amount × weight of this type of assets).

[0079] Transaction heat score = Ʃ (number of transactions on the day ÷ total number of transactions × transaction weight on the day).

[0080] Report score = Ʃ (profit and loss ÷ total profit and loss × profit and loss weight).

[0081] Further, see Figure 6 , 400 including: 410: According to the preference value of each user node in the preference propagation network, a probability model of each user selecting each product is constructed to generate a selection probability matrix of each user selecting each product.

[0082] To ensure that the preference scores are within a reasonable range, StandardScaler is used to standardize the preference scores. This step helps prevent certain features from dominating the clustering results due to their large dimensions.

[0083] The probability of each user choosing each product is modeled, and the softmax function is used to ensure that the sum of the probabilities is 1 and can reflect the relative preferences between different products.

[0084] For example, constructing the user's product selection probability matrix is ​​as follows:

[0085] Each row corresponds to customers u1, u2, u3, and each column corresponds to products p1, p2, p3, p4. Each parameter represents the probability of a user choosing a product. For example, the probability of user 1 choosing product 2 is 0.1112, and the probability of user 3 choosing product 4 is 0.0471.

[0086] 420: Integrate the selection probability matrix and the expanded RFM model parameters to perform hierarchical clustering to assign cluster labels to each user.

[0087] Perform hierarchical clustering using the scipy.cluster.hierarchy.linkage function, selecting the ward method; traverse different cluster numbers from the customized minimum number of clusters (min_clusters) to the customized maximum number of clusters (max_clusters), and calculate the silhouette coefficient (Silhouette Score) for each cluster number. Calculate the silhouette coefficient and select the best number of clusters. Re-perform hierarchical clustering using the best number of clusters and assign the final cluster label to each user.

[0088] Specifically, the silhouette coefficient is an indicator for evaluating the clustering effect, ranging from -1 to 1, and the closer the value is to 1, the better the clustering. The silhouette coefficient function is used to calculate the score for each number of clusters.

[0089] The highest silhouette coefficient and the corresponding number of clusters in each iteration are recorded, and finally the number of clusters with the highest score is selected as the optimal number of clusters.

[0090] Among them, the selection probability matrix is ​​derived from the preference propagation network, which represents the probability of each user choosing different products. The expanded RFM model parameters include indicators such as the last purchase time (Recency), purchase frequency (Frequency), purchase amount (Monetary), as well as expanded transaction heat, asset score, return score and customer comprehensive score parameters.

[0091] 430: Analyze the characteristics of each cluster in the hierarchical clustering to identify typical user behavior patterns.

[0092] Analyze the characteristics of each cluster, identify typical behavior patterns, and provide support for personalized strategies.

[0093] Optionally, the t-SNE technique is used to project the high-dimensional data into three-dimensional space to visually display the distribution of different value groups.

[0094] Optionally, if there is overlap or entanglement, the feature weight distribution and the number of clusters are adjusted, and clustering is performed again.

[0095] Optionally, the clustering method is dynamically adjusted according to the visualization results and business needs to optimize the hierarchical clustering parameters.

[0096] See also Figure 7 , Figure 7 It is a structural diagram of an embodiment of the storage medium provided by the present application.

[0097] The storage medium 500 stores program data 510. When the program data 510 is executed by the processor, the following is achieved: Figure 1 The steps of the distributed processing engine integration method are described.

[0098] The program data 510 is stored in a storage medium 500, and includes a number of instructions for enabling a network device (such as a router, a personal computer, a server, etc.) or a processor to execute all or part of the steps of the methods of the various embodiments of the present application.

[0099] Optionally, the storage medium 500 may be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RFM), a magnetic disk, an optical disk, or other medium that can store program data.

[0100] See also Figure 7 , Figure 7 It is a structural schematic diagram of an embodiment of a computer device provided by the present application.

[0101] The computer device 600 includes a processor 620 and a memory 610 which are connected to each other. The memory 610 stores a computer program. When the processor 620 executes the computer program, the steps of the distributed processing engine integration method described above are implemented.

[0102] Different from the prior art, this application extends the RFM model and introduces new parameters to make the RFM model more suitable for customer segmentation and value assessment scenarios. By constructing an association matrix between users and products, the interaction of each user with each product is reflected. The preference propagation process between users is simulated through a preference propagation network to more accurately predict the user's potential interest in unpurchased products. The extended RFM model is combined with the selection probability matrix to segment users through a clustering algorithm to obtain each user's selection profile for each product, thereby more accurately completing the analysis of the interaction pattern of securities customers.

[0103] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the storage medium embodiment and the computer device embodiment, since they are basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0104] The present application can be used in many general or special computing system environments or configurations, such as personal computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, network PCs, minicomputers, and distributed computing environments including any of the above systems or devices.

[0105] In the several embodiments provided in this application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are only illustrative, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not performed.

[0106] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0107] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0108] The above descriptions are merely embodiments of the present application and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A securities customer analysis method based on the RFM model, characterized in that: include: Obtain asset data and transaction data of securities clients; constructing a preference score matrix between users and products based on the asset data and the transaction data; Constructing a preference propagation network based on the preference score matrix between the user and the product; The target probability is calculated based on the preference value output by the preference propagation network and the extended RFM model; the extended RFM model introduces three factors, namely, transaction heat, asset score and return score, into the asset matching parameters.

2. The securities client analysis method according to claim 1, characterized in that: The constructing a preference score matrix between users and products based on the asset data and transaction data includes: Building a correlation matrix between users and products based on the asset data and the transaction data; Calculate the preference similarity between users based on the association matrix between the users and the products; According to the preference similarities between the users and the actual transaction behaviors of the users, the user's preference for the product is calculated to generate the preference score matrix.

3. The securities client analysis method according to claim 2, characterized in that: The formula for calculating the degree of preference is: Among them, the preference degree (u i , p k ) represents user u i For product k degree of preference, preference similarity (u i ,u j ) represents user u i With user u j The preference similarity between them, transaction amount (u j , p k ) represents user u i For product k The transaction amount.

4. The securities client analysis method according to claim 2, characterized in that: The calculating the user's preference for the product by combining the preference similarity and the user's actual transaction behavior also includes: Obtaining a preset preference weight; the preference weight is set according to business requirements, analysis results of the asset data and the transaction data; The user's preference for the product is adjusted according to the preference weight to generate a weighted preference score matrix.

5. The securities client analysis method according to claim 1, characterized in that: The constructing a preference propagation network based on the preference score matrix between the user and the product includes: Initialize the network structure according to the association matrix between the user and the product and the preference similarity between the users; Setting an initial preference value for each user node in the network structure; A rule for propagating preferences in the network structure is defined based on the initial preference value to form a preference propagation network.

6. The securities client analysis method according to claim 5, characterized in that: The step of defining a rule for propagating preferences in the network structure based on the initial preference value to form a preference propagation network includes: Update the user's preference value for the product through iterative calculation; In response to the preference value satisfying a preset termination condition, the iterative calculation is stopped.

7. The securities client analysis method according to claim 6, characterized in that: The formula for the iterative calculation is: Among them, the new preference value (u i ) indicates that it is used to update user u i The new preference value of the initial preference value, α is the relative importance of controlling the initial preference value and the propagated preference, N(u i ) indicates that the i The set of all adjacent user nodes, N (u j ) indicates that the j The set of all adjacent user nodes, ω(u i ,u j ) represents user u i With user u j The edge weight between j ,u k ) represents user u j With user u k The edge weights between ; Here, α is greater than 0 and less than 1.

8. The securities client analysis method according to claim 1, characterized in that: The calculating the target probability based on the preference value output by the preference propagation network and the extended RFM model includes: According to the preference value of each user node in the preference propagation network, a probability model of each user selecting each product is constructed to generate a selection probability matrix of each user selecting each product; Integrate the selection probability matrix and the expanded RFM model parameters to perform hierarchical clustering to assign cluster labels to each user; The characteristics of each cluster in hierarchical clustering are analyzed to identify typical user behavior patterns.

9. A storage medium having program data stored thereon, characterized in that: When the program data is executed by a processor, the steps of the securities client analysis method as described in any one of claims 1-8 are implemented.

10. A computer device, characterized in that: The method comprises a processor and a memory connected to each other, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the securities client analysis method according to any one of claims 1 to 8 are implemented.

Images