User loss prediction method and device, equipment and storage medium
By combining federated computing and multi-dimensional data, the problem of data silos across different platforms has been solved, enabling accurate prediction of user churn and precise delivery of user activity information, thereby improving the retention rate of bank users.
Patent Information
- Application Number
- CN202311117843.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-31
- Publication Date
- 2025-01-14
AI Technical Summary
Because data is not shared between different platforms, existing technologies that predict user churn based on banks' own data have limited user characteristics, leading to inaccurate predictions and the existence of data silos.
Federated computing is used to process encrypted data from both the payment and banking ends to identify shared user information. This information is then combined with multiple feature dimensions to predict user churn. By analyzing payment data, user payment habits and preferences are analyzed to push precise promotional information.
It improved the accuracy of user churn prediction, solved the data silo problem, and achieved effective integration of bank-side data and billing-side data, thereby improving user retention rate.
Smart Images
Figure CN119313364A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of artificial intelligence technology, and more specifically to a user churn prediction method, device, equipment and storage medium. Background Art
[0002] With the development of financial technology and the rise of third-party payment platforms, the way of paying for daily life has become more convenient. Many users have turned to using mobile applications or third-party payment platforms to pay for daily life instead of relying on traditional banking channels. In order to reduce user churn, banks need to predict the churn of current users and take measures to retain users to minimize corporate losses.
[0003] Due to industry competition, privacy and security issues, data between different platforms are often not interoperable. Therefore, existing technologies usually predict user churn based on the bank's own data, which has problems such as few user characteristics and inaccurate user churn predictions. Summary of the invention
[0004] In view of the above problems, the present disclosure provides a user churn prediction method, apparatus, device and storage medium for improving the accuracy of user churn prediction.
[0005] According to a first aspect of the present disclosure, a user churn prediction method is provided, comprising: receiving first data and second data; wherein the first data is encrypted and uploaded charging-end data, and the second data is encrypted and uploaded bank-end data; performing federated calculations on the first data and the second data to determine user data; wherein the user data includes multiple user features; inputting the user data into a prediction model for prediction to determine a user prediction result; wherein the user prediction result includes a user churn probability and a user target intention; and pushing preset activity information to the user according to the user prediction result.
[0006] According to an embodiment of the present disclosure, before receiving the first data and the second data, it also includes: determining the common user information between the charging end and the bank end based on the encryption sample alignment operation; and determining the data to be uploaded based on the common user information.
[0007] According to an embodiment of the present disclosure, receiving the first data and the second data includes: sending the first public key and the second public key to the charging end and the bank end respectively, so that the charging end and the bank end encrypt the data to be uploaded in the local database based on the first public key and the second public key respectively to obtain the first data and the second data; receiving the first data and the second data uploaded by the charging end and the bank end.
[0008] According to an embodiment of the present disclosure, the shared user information between the charging end and the bank end is determined based on the encrypted sample alignment operation, including: processing the charging end data that has been hash-encrypted based on the first calculation rule to obtain the third data; processing the bank end data that has been hash-encrypted based on the second calculation rule to obtain the fourth data; searching for the same data in the third data and the fourth data; and determining the shared user information based on the same data.
[0009] In some embodiments, a federated computation is performed on the first data and the second data to obtain user data, including: performing a sharding operation on the first data and the second data to obtain a plurality of data slices; combining the data slices based on preset rules to determine a plurality of data slice groups; selecting corresponding functions to perform homomorphic encryption computations on the plurality of data slice groups to obtain a plurality of user features; and obtaining user data of the user based on a set of user features.
[0010] In some embodiments, user data is input into a prediction model for prediction to determine user prediction results, including: performing the following operations for each user: determining multiple user feature vectors of the user based on the user data; performing similarity comparison between the multiple user feature vectors and the portraits of lost users; and determining the user prediction results based on the similarity comparison results.
[0011] In some embodiments, a preset retention activity is pushed to the user based on the user prediction result, including: determining the user's payment preference based on the user's target intention in the user prediction result; and pushing preset activity information to the user based on the user's payment preference.
[0012] A second aspect of the present disclosure provides a user churn prediction device, comprising: a receiving module, used to receive first data and second data; wherein the first data is encrypted and uploaded charging-end data, and the second data is encrypted and uploaded bank-end data; a computing module, used to perform federated computing on the first data and the second data to determine user data; wherein the user data includes multiple user features; a prediction module, used to input user data into a prediction model for prediction, and determine user prediction results; wherein the user prediction results include user churn probability and user target intention; and a push module, used to push preset activity information to the user according to the user prediction results.
[0013] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors execute the above-mentioned user churn prediction method.
[0014] A fourth aspect of the present disclosure further provides a computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the above-mentioned user churn prediction method.
[0015] The disclosure processes the acquired multi-party encrypted data through federated computing to obtain the user data required for user churn prediction, effectively enriches the user data features, improves the accuracy of user churn prediction, solves the data island problem between different platforms, and realizes the effective combination of bank-side data and charging-side data while ensuring the privacy of its own platform data. It predicts user churn based on data of multiple feature dimensions, making user churn prediction more accurate. And because the disclosure obtains charging-side data, it can analyze user payment habits and user payment preferences through charging-side data, thereby realizing accurate push of user activity information, thereby improving the retention rate of bank users. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The above contents and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:
[0017] Figure 1 A diagram schematically shows an application scenario of a user churn prediction method, apparatus, device, and storage medium according to an embodiment of the present disclosure;
[0018] Figure 2 A flowchart of a user churn prediction method according to an embodiment of the present disclosure is schematically shown;
[0019] Figure 3 A flowchart for determining data to be uploaded according to an embodiment of the present disclosure is schematically shown;
[0020] Figure 4 A flowchart for determining data to be uploaded according to an embodiment of the present disclosure is schematically shown;
[0021] Figure 5 A flowchart of receiving first data and second data according to an embodiment of the present disclosure is schematically shown;
[0022] Figure 6 A flowchart of performing federated computing on first data and second data according to an embodiment of the present disclosure is schematically shown;
[0023] Figure 7 Schematically shows a flow chart of performing prediction processing on user data according to an embodiment of the present disclosure;
[0024] Figure 8 A flowchart of pushing a preset retention activity to a user according to an embodiment of the present disclosure is schematically shown;
[0025] Fig. 9 A schematic diagram of a structure of a user churn prediction device according to an embodiment of the present disclosure is shown; and
[0026] Fig.10 A block diagram of an electronic device suitable for implementing a user churn prediction method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION
[0027] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.
[0028] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.
[0029] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.
[0030] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).
[0031] In the technical solution of the present invention, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of relevant data comply with the relevant laws, regulations and standards of relevant countries and regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.
[0032] An embodiment of the present disclosure provides a user churn prediction method, comprising: receiving first data and second data; wherein the first data is encrypted and uploaded charging-end data, and the second data is encrypted and uploaded bank-end data; performing federated calculations on the first data and the second data to determine user data; wherein the user data includes multiple user features; inputting the user data into a prediction model for prediction to determine a user prediction result; wherein the user prediction result includes a user churn probability and a user target intention; and pushing preset activity information to the user based on the user prediction result.
[0033] Figure 1 The following schematically shows an application scenario diagram of the user churn prediction method according to an embodiment of the present disclosure.
[0034] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a terminal device 101, a terminal device 102, a terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.
[0035] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).
[0036] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.
[0037] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.
[0038] It should be noted that the user churn prediction method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the user churn prediction device provided in the embodiment of the present disclosure can generally be set in the server 105. The user churn prediction method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the user churn prediction device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.
[0039] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.
[0040] The following will be based on Figure 1 The scene described by Figure 2 to Figure 8 The user churn prediction method of the disclosed embodiment is described in detail.
[0041] Figure 2 The flowchart of the user churn prediction method according to an embodiment of the present disclosure is schematically shown.
[0042] like Figure 2 As shown, the user churn prediction method of this embodiment includes operations S210 to S240.
[0043] In operation S210, first data and second data are received; wherein the first data is encrypted and uploaded charging-end data, and the second data is encrypted and uploaded bank-end data.
[0044] In some embodiments, the data of different platforms reflect different user characteristics. For example, the data on the charging end mainly reflects the user's payment habits, while the data on the bank end often includes user information in the bank. Due to industry competition, privacy and security issues, data on different platforms are usually not shared, and the data on each platform usually exists in the form of isolated islands. The first data and the second data received by the present disclosure are both uploaded by different platforms after encrypting the data locally, which can effectively prevent the problem of leakage of original data.
[0045] In operation S220, a federated calculation is performed on the first data and the second data to obtain user data; wherein the user data includes a plurality of user features.
[0046] In some embodiments, a federated calculation is performed on the encrypted first data and the second data, and user data including the charging end dimension and the bank end dimension is obtained through the federated calculation. Compared with user churn prediction using only the user data of the bank dimension, the dimensions in the user data of the present disclosure are more comprehensive. Among them, performing a federated calculation on the first data and the second data may include: first decrypting the encrypted first data and the second data, and determining the user data based on the decrypted first data and the second data. Or, directly calculating the encrypted first data and the second data to obtain the encrypted user data, and performing a decryption operation on the encrypted user data to obtain the user data.
[0047] In the specific implementation process, the user data obtained after federal calculation may include, for example: user payment frequency, payment amount, payment history, payment item type, payment channel, payment method, user attribute information, etc. Among them, user payment frequency refers to the average number of payments made by the user in each cycle; payment history refers to the payment information of the user in the past period of time (such as the payment status of the user in the past three months); payment item type refers to the specific items paid by the user, such as living payment, shopping payment, entertainment payment, government payment, etc.; payment channel refers to the payment method used by the user, such as payment software, online banking, mobile banking, etc.; payment method refers to the payment method used by the user when paying, such as Alipay, WeChat, bank card, other bank card, etc.; user attribute information refers to user personal information, such as age, gender, occupation, number of years as a user of the bank, etc.
[0048] In operation S230, the user data is passed into the prediction model to perform prediction processing to determine the user prediction result; wherein the user prediction result includes the user churn probability and the user target intention.
[0049] In some embodiments, the user data obtained after the federated calculation is passed to the prediction model for prediction. The prediction model may be, for example, a logistic regression model, which calculates user features based on the user data and determines the user prediction results based on the user features. The user prediction results include not only the user churn probability, but also the user's target intention, that is, the target platform that the user is more likely to choose.
[0050] In operation S240, preset activity information is pushed to the user according to the user prediction result.
[0051] In some embodiments, preset activity information is pushed to the user based on the target intention in the user prediction result, so as to achieve the purpose of improving the user retention rate.
[0052] The disclosure processes the obtained multi-party encrypted data through federated computing to obtain the user data required for user churn prediction, effectively enriches the user data features, improves the accuracy of user churn prediction, solves the data island problem between different platforms, and realizes the effective combination of bank-side data and charging-side data while ensuring the privacy of its own platform data. It predicts user churn based on data of multiple feature dimensions, making user churn prediction more accurate. And because the disclosure obtains charging-side data, it can use the charging-side data to obtain user payment habits and user payment preferences, thereby realizing the accurate push of user activity information, thereby improving the retention rate of bank users.
[0053] Figure 3 The flowchart of determining data to be uploaded according to an embodiment of the present disclosure is schematically shown.
[0054] like Figure 3 As shown, the process of determining the data to be uploaded in this embodiment includes operations S310 to S320.
[0055] In operation S310, it is determined that the charging end and the bank end have common user information based on the encrypted sample alignment operation.
[0056] In operation S320, data to be uploaded is determined according to the shared user information.
[0057] In some embodiments, there is more overlap in user space and less overlap in data feature space between the charging end and the bank end. For example, the users between the bank end and the charging end are relatively similar, but due to the large differences in the business types between the bank end and the charging end, there are also large differences between the data features recorded by the bank end and the charging end. Among them, the data on the bank end are mostly user characteristics, such as the user's age, gender, occupation, etc., and the data on the charging end are mostly payment characteristics, such as the user's payment habits, payment frequency, etc.
[0058] Therefore, the present disclosure proposes, firstly, based on the encryption alignment operation, to screen out the common user information of the bank side and the charging side without leaking the original data of each side, determine the data to be uploaded according to the common user information, upload the data after processing, and perform federal computing based on the data uploaded by the bank side and the charging side, so as to effectively save server resources, eliminate irrelevant data, and speed up the data processing rate.
[0059] In the specific implementation process, the mobile phone number can be used as a parameter to determine the shared user information to implement the screening of shared user information in the bank end and the charging end. For example, the encrypted bank end user number and the charging end user number can be obtained respectively, and the same user numbers in the bank end and the charging end can be compared through sample alignment technology to determine the shared user information. The shared user information is sent to the charging end of the bank end respectively, so that the bank end and the charging end determine the data to be uploaded based on the shared user information.
[0060] Figure 4 The flowchart of determining data to be uploaded according to an embodiment of the present disclosure is schematically shown.
[0061] like Figure 4 As shown, the process of determining the data to be uploaded in this embodiment includes operations S410 to S440.
[0062] In operation S410, the charging terminal data that has been hash-encrypted is processed based on the first calculation rule to obtain third data.
[0063] In some embodiments, the first calculation rule includes a first calculation formula and a second calculation formula. First, the first calculation formula is used to process the charging terminal data that has been hashed. The specific steps include: first, using the parameter R i A multiplication operation is performed on the charging terminal data M that has been hashed to obtain data M1, and then a power operation is performed on the data M1 using the parameter d to obtain data M2.
[0064] After operation S420 is completed, the data M2 is processed using the second calculation formula. The specific steps include: first, using the parameter R i A division operation is performed on the data M2 to obtain data M3, and then a hash encryption operation is performed on the data M3 to obtain third data M4. At this time, the charging end data M is a collection of charging end user numbers.
[0065] In operation S420, the bank-side data that has been hash-encrypted is processed based on the second calculation rule to obtain fourth data.
[0066] In some embodiments, the calculation is performed on the bank-side data that has been hashed based on the second calculation rule, including: firstly, using the parameter d to perform a power calculation operation on the bank-side data N that has been hashed to obtain data N1, and then performing a hash operation on the data N1 to obtain fourth data N2. At this time, the bank-side data N is a set of bank-side user numbers.
[0067] In operation S430, identical data in the third data and the fourth data is searched.
[0068] In operation S440, common user information is determined based on the same data.
[0069] In some embodiments, the bank-side data and the charging-side data are processed respectively by the first calculation rule and the second calculation rule, so that the same data in the encrypted bank-side data and the charging-side data have the same presentation form, and the data with the same presentation form is a shared user. The comparison result (i.e., shared user information) is sent to the charging end and the bank end respectively, so that the bank end and the charging end determine the data to be uploaded based on the shared user information.
[0070] In addition to the users of this bank, there are many other users in the charging end. In the process of user churn prediction and analysis, it is only necessary to obtain the charging end data features related to the bank's predicted users, without analyzing all the user data of the charging end. In order to effectively save the computing resources of the server and improve the efficiency of user churn prediction, the present disclosure proposes to first obtain the bank end and the charging end data to obtain the shared user information and realize the screening of shared users.
[0071] In order to determine the common users on the bank side and the charging side, it is necessary to put the bank side data and the charging side data together for intersection processing. The present disclosure proposes to use the first calculation rule and the second calculation rule to process the bank side data and the charging side data, so that the bank side data and the charging side data are still the same after encryption operation, so as to achieve sample alignment without leaking data privacy and determine the common user information of the charging side and the bank side.
[0072] Figure 5 The flowchart of receiving first data and second data according to an embodiment of the present disclosure is schematically shown.
[0073] like Figure 5 As shown, the receiving of the first data and the second data in this embodiment includes operations S510 to S520.
[0074] In operation S510, the first public key and the second public key are sent to the charging end and the bank end respectively, so that the charging end and the bank end encrypt the data to be uploaded in the local database based on the first public key and the second public key respectively to obtain the first data and the second data.
[0075] In operation S520, first data and second data uploaded by the charging end and the bank end are received.
[0076] In some embodiments, the server generates a public key-private key pair for encrypting the first data and the second data, wherein the public key is a public key, which is published by the key pair holder and used to encrypt the data; the private key is a private key, which is held by the key pair holder and is not published to the public, and is used to decrypt the data encrypted by the public key.
[0077] The server sends the first public key and the second public key to the charging end and the bank end respectively. The charging end encrypts the data to be uploaded in the local database based on the received first public key to obtain the first data; the bank end encrypts the data to be uploaded in the local database based on the received second public key to obtain the second data.
[0078] The charging end and the bank end encrypt the data to be uploaded based on the first public key and the second public key before uploading. In addition to ensuring that their own data privacy is not leaked, it can also effectively ensure the security of data transmission. Even if the first data and the second data are stolen by a third party during the data uploading process, the first data and the second data cannot be cracked, thereby ensuring data security.
[0079] During the specific implementation process, when the server sends the first public key and the second public key to the charging end and the bank end, it also needs to send a digital signature to the charging end and the bank end. The charging end and the bank end use the digital signature to verify the authenticity of the received first public key and the second public key to ensure that the first public key and the second public key are the correct keys sent by the legitimate server, so as to avoid the risk of data leakage due to receiving false public keys sent by illegal servers, and further ensure data security.
[0080] Figure 6 The flowchart of performing federated computing on first data and second data according to an embodiment of the present disclosure is schematically shown.
[0081] like Figure 6 As shown, performing federated computing on first data and second data in this embodiment includes operations S610 to S640.
[0082] In operation S610, a slicing operation is performed on first data and second data to obtain a plurality of data slices.
[0083] In operation S620, the data slices are combined based on a preset rule to determine a plurality of data slice groups.
[0084] In operation S630, a corresponding function is selected to perform homomorphic encryption calculations on multiple data slice groups to obtain multiple user features.
[0085] In operation S640, user data of the user is obtained based on the set of user characteristics.
[0086] In some embodiments, by performing a sharding operation on the first data and the second data, data slices of different data types are obtained, and then the data slices are combined based on a preset rule to obtain multiple data groups, wherein each data group will reflect different user characteristics after calculation. By combining different types of data slices, user characteristics can be expanded, making the user characteristics after homomorphic encryption calculation richer.
[0087] In the specific implementation process, the data slice is calculated using a fully homomorphic encryption algorithm, and the calculation is performed when both the first data and the second data are in an encrypted state to obtain the encrypted user features, and the encrypted user features are decrypted using a private key to obtain the user features. At this time, compared with the original data on the charging end and the bank end, the user features are calculated data, and the original data is desensitized. The server can only obtain the user features it wants, and cannot view the original data on the charging end and the bank end, thus achieving the enrichment of user features while protecting data privacy.
[0088] Figure 7 The flowchart of performing prediction processing on user data according to an embodiment of the present disclosure is schematically shown.
[0089] like Figure 7 As shown, the prediction process performed on the user data in this embodiment includes operations S710 to S730.
[0090] In operation S710, a plurality of user feature vectors of the user are determined based on user data.
[0091] In operation S720, a plurality of user feature vectors are compared with the lost user portraits for similarity.
[0092] In operation S730, based on the similarity comparison result, a user prediction result is determined.
[0093] In some embodiments, the user data obtained after the federated calculation is passed into the prediction model, and the prediction model is used to perform prediction processing on the user data to determine the user prediction result.
[0094] User data is processed by word embedding and converted into feature vectors. These data are fused in the fusion layer of the prediction model and then input into the classification layer for classification to obtain user prediction results. Classification in the classification layer includes: comparing the similarity between the user's feature vector and the lost user portrait, and taking the lost user portrait corresponding to the highest similarity as the user prediction result. Among them, the user prediction result includes the user's churn probability and the user's target intention (that is, the target platform that the user is more likely to choose). The lost user portrait is drawn by the prediction model based on the lost user data.
[0095] In the specific implementation process, the prediction model is constructed as follows:
[0096] h(x)=1 / (1+exp(-z))
[0097] Among them, h(x) is the user churn probability, z is the linear combination of user characteristics and model parameters, which can be expressed as: z = θ0 + θ1x1 + θ2x2 + … + θ n x n , among them, θ0, θ1, θ2...θ n are the parameters of the model, x1, x2, ...x n is the eigenvalue.
[0098] Figure 8 The flowchart of pushing a preset retention activity to a user according to an embodiment of the present disclosure is schematically shown.
[0099] like Figure 8 As shown, the pushing of a preset retention activity to a user in this embodiment includes operations S810 to S820.
[0100] In operation S810, a user payment preference is determined based on a user goal intention in a user prediction result.
[0101] In operation S820, preset activity information is pushed to the user based on the user's payment preference.
[0102] In some embodiments, different target platforms have different rules. Therefore, the user's payment preference can be inferred based on the user's target intention, and corresponding activity information can be pushed to the user based on the user's payment preference. Accurate push of activity information based on the user's payment preference can effectively improve the user's retention rate.
[0103] In addition, the payment preferences of users with different characteristics can be analyzed based on the first data and the second data. For example, the payment preferences of female users and male users can be analyzed to achieve personalized push notifications for users of different genders, and different activity information can be pushed to female users and male users respectively. For another example, the payment preferences of users of different age groups can be analyzed based on user age groups, and personalized activity information can be pushed to users of different age groups. This method of pushing personalized activity information to users with different characteristics can effectively improve user stickiness and retention rates.
[0104] The user churn prediction method provided by the present disclosure obtains data from multiple platforms through federated computing, effectively solves the island problem between data from different platforms, realizes the effective combination of data from different platforms while protecting data privacy, uses multi-dimensional data to predict user churn, and realizes the effective combination of bank-side data and fee-side data to obtain richer bank user characteristics, effectively improving the accuracy of user churn prediction. In addition to predicting the probability of user churn, the user churn prediction method provided by the present disclosure can also predict the target intention of users after churn, and flexibly select activity information based on the user's target intention, so as to realize the accurate push of activity information, thereby effectively improving the user retention rate.
[0105] Based on the above user churn prediction method, the present disclosure also provides a user churn prediction device. Fig. 9 The device is described in detail.
[0106] Fig. 9 The structural block diagram of the user churn prediction device according to an embodiment of the present disclosure is schematically shown.
[0107] like Fig. 9 As shown, the user churn prediction device 900 of this embodiment includes a receiving module 910 , a calculating module 920 , a predicting module 930 and a pushing module 940 .
[0108] The receiving module 910 is used to receive the first data and the second data, wherein the first data is the encrypted uploaded charging end data, and the second data is the encrypted uploaded bank end data. In one embodiment, the receiving module 910 can be used to perform the operation S210 described above, which will not be described in detail here.
[0109] The calculation module 920 is used to perform federated calculation on the first data and the second data to determine user data, wherein the user data includes multiple user features. In one embodiment, the calculation module 920 can be used to perform the operation S220 described above, which will not be repeated here.
[0110] The prediction module 930 is used to input user data into the prediction model for prediction and determine the user prediction result; wherein the user prediction result includes the user churn probability and the user target intention. In one embodiment, the prediction module 930 can be used to perform the operation S230 described above, which will not be repeated here.
[0111] The push module 940 is used to push the preset activity information to the user according to the user prediction result. In one embodiment, the push module 940 can be used to perform the operation S240 described above, which will not be repeated here.
[0112] According to an embodiment of the present disclosure, any multiple modules of the receiving module 910, the computing module 920, the prediction module 930, and the push module 940 can be combined in one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the receiving module 910, the computing module 920, the prediction module 930, and the push module 940 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware, and firmware or in any appropriate combination of any of them. Alternatively, at least one of the receiving module 910, the computing module 920, the prediction module 930, and the push module 940 can be at least partially implemented as a computer program module, and when the computer program module is run, the corresponding function can be performed.
[0113] Fig.10 A block diagram of an electronic device suitable for implementing a user churn prediction method according to an embodiment of the present disclosure is schematically shown.
[0114] like Fig.10 As shown, the electronic device 1000 according to an embodiment of the present disclosure includes a processor 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage part 1009 into a random access memory (RAM) 1003. The processor 1001 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 1001 may also include an onboard memory for caching purposes. The processor 1001 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.
[0115] In RAM 1003, various programs and data required for the operation of electronic device 1000 are stored. Processor 1001, ROM 1002 and RAM 1003 are connected to each other via bus 1004. Processor 1001 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 1002 and / or RAM 1003. It should be noted that the program can also be stored in one or more memories other than ROM 1002 and RAM 1003. Processor 1001 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.
[0116] According to an embodiment of the present disclosure, the electronic device 1000 may further include an input / output (I / O) interface 1005, which is also connected to the bus 1004. The electronic device 1000 may further include one or more of the following components connected to the I / O interface 1005: an input portion 1006 including a keyboard, a mouse, etc.; an output portion 1007 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 1009 including a hard disk, etc.; and a communication portion 1009 including a network interface card such as a LAN card, a modem, etc. The communication portion 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the I / O interface 1005 as needed. A removable medium 1011, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1010 as needed, so that a computer program read therefrom is installed into the storage portion 1009 as needed.
[0117] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.
[0118] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 1002 and / or RAM 1003 described above and / or one or more memories other than ROM 1002 and RAM 1003.
[0119] The flow chart and block diagram in the accompanying drawings illustrate the possible implementation architecture, function and operation of the system and method according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0120] It will be appreciated by those skilled in the art that the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations and / or combinations are not explicitly described in the present disclosure. In particular, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways without departing from the spirit and teachings of the present disclosure. All of these combinations and / or combinations fall within the scope of the present disclosure.
[0121] The embodiments of the present disclosure are described above. However, these embodiments are only for the purpose of illustration and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present disclosure.
Claims
1. A user churn prediction method, comprising: Receive first data and second data; wherein the first data is encrypted and uploaded charging end data, and the second data is encrypted and uploaded bank end data; Performing federated computation on the first data and the second data to determine user data; wherein the user data includes a plurality of user features; Inputting the user data into a prediction model for prediction to determine a user prediction result; wherein the user prediction result includes a user churn probability and a user target intention; Preset activity information is pushed to the user according to the user prediction result.
2. The user churn prediction method according to claim 1, before receiving the first data and the second data, further comprising: Determine the shared user information between the charging end and the bank end based on the encrypted sample alignment operation; The data to be uploaded is determined according to the shared user information.
3. The user churn prediction method according to claim 2, wherein the receiving the first data and the second data comprises: Sending the first public key and the second public key to the charging end and the bank end respectively, so that the charging end and the bank end encrypt the data to be uploaded in the local database based on the first public key and the second public key respectively, to obtain the first data and the second data; The first data and the second data uploaded by the charging end and the bank end are received.
4. The user churn prediction method according to claim 3, wherein the step of determining that the charging end and the bank end share user information based on the encrypted sample alignment operation comprises: Processing the charging terminal data after hash encryption based on the first calculation rule to obtain third data; Processing the bank-side data after hash encryption based on the second calculation rule to obtain fourth data; searching for identical data in the third data and the fourth data; The common user information is determined based on the same data.
5. The user churn prediction method according to claim 1, wherein the step of performing a federated calculation on the first data and the second data to determine the user data comprises: Performing a sharding operation on the first data and the second data to obtain a plurality of data slices; Combining the data slices based on a preset rule to determine a plurality of data slice groups; Select a corresponding function to perform homomorphic encryption calculation on the multiple data slice groups to obtain multiple user features; Based on the set of user characteristics, user data of the user is determined.
6. The user churn prediction method according to claim 1, wherein the step of inputting the user data into a prediction model for prediction and determining a user prediction result comprises: For each user, do the following: Determine a plurality of user feature vectors of the user based on the user data; Comparing the multiple user feature vectors with the lost user portraits for similarity; Based on the similarity comparison result, a user prediction result is determined.
7. The user churn prediction method according to claim 1, wherein the step of pushing preset activity information to the user according to the user prediction result comprises: Determine the user's payment preference based on the user's target intention in the user prediction results; Based on the user's payment preference, preset activity information is pushed to the user.
8. A user churn prediction device, comprising: A receiving module, configured to receive first data and second data; wherein the first data is encrypted and uploaded charging terminal data, and the second data is encrypted and uploaded bank terminal data; A computing module, configured to perform a federated computing on the first data and the second data to determine user data; wherein the user data includes a plurality of user features; A prediction module, used to input the user data into a prediction model for prediction and determine a user prediction result; wherein the user prediction result includes a user churn probability and a user target intention; and The push module is used to push preset activity information to the user according to the user prediction result.
9. An electronic device, comprising: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors are enabled to execute the method according to any one of claims 1 to 7.
10. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, causes the processor to execute the method according to any one of claims 1 to 7.