Bank User Risk Early Warning Method and Device Based on Big Data

Abstract

The present application discloses a big-data-based bank user risk early warning method and device, which relates to the financial field. The method includes: receiving a service handling request submitted by a user; obtaining dynamic data of the user authorized by the user and a third-party institution according to the service handling request; performing data cleaning and data governance on the dynamic data to obtain evaluation data of the user; determining a risk evaluation result of the user according to the evaluation data and a pre-created risk early warning model; wherein the risk early warning model is pre-trained according to the dynamic data of multiple users and corresponding ones. The present application performs user risk assessment based on the dynamic data of the user, solves the shortcoming problem of the existing dynamic assessment method for bank customer risk early warning, and provides a more dynamic, vivid, real-time and highly valuable, brand-new way for user assessment in the process of bank service handling.

Description

Technical Field

[0001] This application relates to the technical field of finance, and particularly to a method and device for early warning of bank user risks based on big data. Background Art

[0002] Currently, with the increasing improvement of China's Internet infrastructure and the accelerating popularization of mobile terminals, China is accelerating into the digital age. E-commerce and e-shopping have gradually become the mainstream of shopping. The digitization of many scattered, unstandardized and digital economic behaviors of individuals and units in the past has become the norm, and the online acceleration of economic behaviors in the Internet era has accelerated. This also makes the digital information of individuals and units an important data source for banks to predict customer risks.

[0003] Traditional bank customer risk early warning has defects such as a single means and inability to perceive in real time. At present, the bank's user risk early warning assessment mainly adopts the tracking and assessment of traditional economic behaviors such as user repayment, overdue, and cash flow. However, these data have relatively low reliability and a single means. Since the data are all feedback after the fact or even lagged, they cannot dynamically reflect the true credit status of bank customers. Summary of the Invention

[0004] In order to solve the problems existing in the prior art, on the first hand, this application provides a method for early warning of bank user risks based on big data, which relates to the financial field. The method includes:

[0005] Receiving a business handling request submitted by a user;

[0006] Obtaining dynamic data of the user authorized by the user and a third-party institution according to the business handling request, where the dynamic data includes work and rest data and express delivery data;

[0007] Performing data cleaning and data governance on the dynamic data to obtain evaluation data of the user;

[0008] Determining a risk assessment result of the user according to the evaluation data and a pre-created risk early warning model; wherein, the risk early warning model is pre-trained according to the dynamic data of multiple users and corresponding ones.

[0009] In an embodiment, the obtaining of the dynamic data of the user authorized by the user and a third-party institution according to the business handling request includes:

[0010] Obtaining the work and rest data of the user; the work and rest data includes the user's working hours, off-duty hours, work address, and residential address; the work and rest data is obtained under the authorization of the user;

[0011] Obtain the user's express delivery data from multiple express delivery companies; the express delivery data includes the express delivery company brand, recipient address, sender address, recipient information, sender information, inner item category, sending time, receiving time, and receiving method; the express delivery data is obtained under the authorization of the user and the express delivery company.

[0012] In one embodiment, the data cleaning and data governance of the dynamic data to obtain the user's evaluation data includes:

[0013] Filter the dynamic data according to a preset evaluation time range; and

[0014] Convert the filtered dynamic data into the preset data format to obtain the evaluation data.

[0015] In one embodiment, the steps of creating the risk warning model include:

[0016] Obtain multiple sample dynamic data authorized by the user himself, and the dynamic data includes sample work and rest data and sample express delivery data;

[0017] Generate a training data set according to the multiple sample dynamic data;

[0018] Use the training data set to train a pre-established machine learning model to obtain the risk warning model.

[0019] In one embodiment, the generating the training data set according to the multiple sample dynamic data includes:

[0020] Perform data cleaning and data governance on the multiple sample dynamic data to obtain initial data;

[0021] Determine the classification labels of the initial data according to preset label rules;

[0022] Generate the training data set according to the initial data and its corresponding classification labels.

[0023] In one embodiment, the determining the classification labels of the initial data according to preset label rules includes:

[0024] Obtain the sending time, receiving time, working time, and off-duty time of the same user in the initial data;

[0025] Determine the high-frequency sending and receiving time periods of the user according to the sending time and the receiving time;

[0026] Determine the working time period of the user according to the working time and the off-duty time;

[0027] Determine the classification labels according to the high-frequency sending and receiving time periods and the working time period.

[0028] In one embodiment, determining the classification label according to the high-frequency collection and delivery time period and the working time period includes:

[0029] Determining a first score according to the matching degree between the high-frequency collection and delivery time period and the working time period and a preset scoring table;

[0030] Determining a second score according to the stability degree of the high-frequency collection and delivery time period and the scoring table;

[0031] Determining a third score according to the matching degree between the high-frequency collection and delivery time period and a preset abnormal time period and the scoring table;

[0032] Determining the classification label according to the sum of the first score, the second score, and the third score and a preset mapping relationship between the classification label and the score.

[0033] In a second aspect, the present application further provides a big data-based bank user risk warning device, including:

[0034] A request receiving module, configured to receive a service handling request submitted by a user;

[0035] A dynamic data acquisition module, configured to acquire dynamic data of a user authorized by the user and a third-party institution according to the service handling request, where the dynamic data includes work and rest data and express delivery data;

[0036] A data preprocessing module, configured to perform data cleaning and data governance on the dynamic data to obtain evaluation data of the user;

[0037] A risk assessment module, configured to determine a risk assessment result of the user according to the evaluation data and a pre-created risk warning model; wherein, the risk warning model is pre-trained according to dynamic data of multiple users and corresponding ones.

[0038] In one embodiment, the dynamic data acquisition module includes:

[0039] A work and rest data acquisition unit, configured to acquire work and rest data of the user; the work and rest data includes the user's work start time, work end time, work address, and residential address; the work and rest data is acquired under the condition of user authorization;

[0040] An express delivery data acquisition unit, configured to acquire express delivery data of the user from multiple express delivery companies; the express delivery data includes the express delivery company brand, recipient address, sender address, recipient information, sender information, internal item category, sending time, receiving time, and receiving method; the express delivery data is acquired under the condition of user and express delivery company authorization.

[0041] In one embodiment, the data preprocessing module is specifically configured to:

[0042] Filter the dynamic data according to a preset evaluation time range; and

[0043] Convert the filtered dynamic data into the evaluation data according to a predefined data format.

[0044] In one embodiment, the big data-based bank user risk warning device further includes:

[0045] A sample data acquisition module, configured to acquire a plurality of sample dynamic data authorized by the user himself / herself, where the dynamic data includes sample work and rest data and sample express delivery data;

[0046] A training data generation module, configured to generate a training data set according to the plurality of sample dynamic data;

[0047] A model training module, configured to train a pre-established machine learning model using the training data set to obtain the risk warning model.

[0048] In one embodiment, the training data generation module includes:

[0049] A data preprocessing unit, configured to perform data cleaning and data governance on the plurality of sample dynamic data to obtain initial data;

[0050] A classification label determination unit, configured to determine the classification label of the initial data according to a preset label rule;

[0051] A training data generation unit, configured to generate the training data set according to the initial data and its corresponding classification label.

[0052] In one embodiment, the classification label determination unit includes:

[0053] A concerned data acquisition subunit, configured to acquire the sender time, recipient time, work start time, and work end time of the same user in the initial data;

[0054] A high-frequency sending and receiving period acquisition subunit, configured to determine the high-frequency sending and receiving period of the user according to the sender time and the recipient time;

[0055] A work period acquisition subunit, configured to determine the work period of the user according to the work start time and the work end time;

[0056] A classification label determination subunit, configured to determine the classification label according to the high-frequency sending and receiving period and the work period.

[0057] In one embodiment, the classification label determination subunit is specifically configured to:

[0058] Determine a first score according to the matching degree between the high-frequency sending and receiving time periods and the working time period and a preset scoring table.

[0059] Determine a second score according to the stability degree of the high-frequency sending and receiving time periods and the scoring table.

[0060] Determine a third score according to the matching degree between the high-frequency sending and receiving time periods and a preset abnormal time period and the scoring table.

[0061] Determine the classification label according to the sum of the first score, the second score, and the third score and the mapping relationship between the preset classification label and the score.

[0062] In a third aspect, the present application provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, any bank user risk warning method provided by the present application is implemented.

[0063] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, any bank user risk warning method provided by the present application is implemented.

[0064] The present application conducts user risk assessment based on the dynamic data of users, and more specifically, combines the express delivery data and work and rest data of users for assessment, solving the shortcoming problem of the existing dynamic assessment method for bank customer risk warning, and providing a more dynamic, vivid, real-time, high-value, and brand-new method for user assessment in the process of bank business handling. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. In the drawings:

[0066] Figure 1 It is a schematic diagram of the bank user risk warning method based on big data provided by the present application.

[0067] Figure 2 It is a schematic diagram of the steps for obtaining user dynamic data provided by the present application.

[0068] Figure 3 It is a schematic diagram of the steps for preprocessing the dynamic data of users provided by the present application.

[0069] Figure 4 Schematic diagram of steps for creating a risk warning model provided by this application.

[0070] Figure 5 Schematic diagram of steps for generating a training data set provided by this application.

[0071] Figure 6 Schematic diagram of steps for determining classification labels provided by this application.

[0072] Figure 7 Schematic diagram of steps for creating a risk warning model provided by this application.

[0073] Figure 8 Schematic diagram of a risk warning device for bank users based on big data provided by this application.

[0074] Figure 9 Another schematic diagram of a risk warning device for bank users based on big data provided by this application.

[0075] Figure 10 Another schematic diagram of a risk warning device for bank users based on big data provided by this application.

[0076] Figure 11 Another schematic diagram of a risk warning device for bank users based on big data provided by this application.

[0077] Figure 12 Schematic diagram of a computer device provided by this application. Detailed implementation manners

[0078] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer and more understandable, the following further describes the embodiments of the present invention in detail with reference to the accompanying drawings. Herein, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but not to limit the present invention.

[0079] It should be noted that the risk warning method and device for bank users based on big data of this application can be used in the financial field and can also be used in any field other than the financial field. This application does not limit the application fields of the risk warning method and device for bank users based on big data.

[0080] In the technical solutions of this application, the acquisition, storage, use, processing, etc. of data all comply with the relevant regulations of national laws and regulations. The user information in each embodiment is obtained through legal and compliant channels, and the acquisition, storage, use, processing, etc. of user information have obtained the authorization and consent of the customers.

[0081] To solve the problems existing in the prior art, in a first aspect, the present application provides a method for early warning of bank user risks based on big data, which relates to the financial field. As Figure 1 shown, the method includes steps S101 to S104:

[0082] Step S101, receiving a business handling request submitted by a user.

[0083] Specifically, the business handling request of the user includes, but is not limited to, a bank loan business handling request and other businesses that need to be handled after evaluating the user's credit information.

[0084] Step S102, obtaining dynamic data of the user authorized by the user and a third-party institution according to the business handling request.

[0085] Specifically, the dynamic data here includes the user's work and rest data and express delivery data. The work and rest data includes, but is not limited to, the user's work address, residential address, daily work start time, work end time, etc. These data can be provided by the user with the user's authorization. To ensure the accuracy and authenticity of the data, the work start time and work end time (such as commuting clock-in records) of the user can be provided by the user's work unit with the authorization of the user and the work unit at the same time.

[0086] The express delivery data includes, but is not limited to, the sending time, sending address, sender information, express delivery company brand, internal item category when the user is the sender, and the receiving time, receiving method, receiving address, recipient information, express delivery company brand, internal item category, etc. when the user is the recipient. Since the express delivery data is usually retained by each express delivery company, the express delivery data needs to be obtained with the authorization of both the user himself and the express delivery company.

[0087] The third-party institutions referred to in this step mainly refer to the user's work unit and express delivery company. The purpose of cooperating with the third-party institutions is to ensure the authenticity and reliability of the dynamic data of the user obtained, so as to improve the accuracy of the risk assessment result of the present application.

[0088] Step S103, performing data cleaning and data governance on the dynamic data to obtain the evaluation data of the user.

[0089] This step is to screen and convert the format of the dynamic data obtained in step S102. On the one hand, part of the data that is not used as model input data in the dynamic data is removed, and on the other hand, the data that the user needs to input into the model is converted into a unified format to facilitate the model to analyze and identify.

[0090] Step S104: Determine the risk assessment result of the user according to the evaluation data and a pre-created risk warning model; wherein, the risk warning model is obtained by pre-training based on the dynamic data of multiple users and corresponding ones.

[0091] Specifically, input the evaluation data obtained in step S103 into the risk warning model, so that the risk warning model performs feature analysis on the evaluation data and outputs the corresponding risk assessment result. The steps of creating the risk warning model will be described in detail in the subsequent embodiments.

[0092] In this embodiment, the work and rest data and express delivery data in the user's daily life are combined as the basis for user risk assessment. The data in the user's daily life can truly and timely reflect the changes in the user's daily economic behavior. Compared with the existing tracking and assessment of traditional economic behaviors such as the user's repayment, overdue, and cash flow, the daily economic behavior can dynamically and timely reflect the user's true credit status, and has higher real-time performance, and it is not easy to have the situation of lagged feedback of traditional economic behaviors. This application can either conduct user risk assessment alone or combine with existing other means to jointly conduct user risk assessment.

[0093] In one embodiment, as Figure 2 shown, step S102: Obtain the dynamic data of the user authorized by the user and the third party according to the service handling request, including:

[0094] Step S1021: Obtain the work and rest data of the user; the work and rest data includes the user's work start time, work end time, work address, and residential address; the work and rest data is obtained under the condition of the user's authorization.

[0095] Step S1022: Obtain the express delivery data of the user from multiple express delivery companies; the express delivery data includes the express delivery company brand, recipient address, sender address, recipient information, sender information, internal item category, sending time, receiving time, and receiving method; the express delivery data is obtained under the condition of the user's and the express delivery company's authorization.

[0096] This embodiment describes the types of dynamic data and the acquisition methods of dynamic data involved in step S102. Since it may be necessary to cooperate with third-party institutions such as the user's work unit and the express delivery companies the user has cooperated with to obtain data, in order to strengthen the protection of user privacy, the third-party institution can first perform de-encryption processing before providing the above dynamic data, and encrypted transmission should also be used during the data transmission process. Further, the third-party institution can directly interact with the bank, or can elect an authoritative non-profit platform as an intermediary platform to realize the interaction between the bank and the third-party institution.

[0097] In one embodiment, as Figure 3As shown in the figure, in step S103, data cleaning and data governance are performed on the dynamic data to obtain the evaluation data of the user, including:

[0098] Step S1031, screening the dynamic data according to a preset evaluation time range.

[0099] Specifically, in order to ensure the objectivity of the dynamic data, it is necessary to obtain the user's recent dynamic data. The screening in the time dimension can be achieved through a preset evaluation time range. The evaluation time range can be, for example, the last 3 years, the last 1 year, etc., and can be set according to actual needs.

[0100] In addition to the screening in the time dimension, further screening of the data validity is also required to remove invalid data. For example, for express delivery data, the dynamic data can be screened according to the receiving method, and the dynamic data with the receiving method of signed by the recipient himself / herself is screened out, and the dynamic data with the receiving method of received on behalf of others is excluded. Since the time when the user sends and receives express deliveries in this application is key information, it is necessary to retain as much information as possible that can reflect the time pattern of the user sending and receiving express deliveries by himself / herself.

[0101] Step S1032, converting the screened dynamic data according to a predefined data format to obtain the evaluation data.

[0102] Specifically, the obtained dynamic data is converted according to a predefined data format. For example, for the inner item category in the express delivery data, according to the data provided by the express company, there are 10 categories of inner item categories, while the bank side may divide the inner item categories into 6 categories. At this time, the express delivery data provided by the express company needs to be re-divided according to the corresponding relationship of the inner item categories, that is, the original 10 categories of the express company are correspondingly replaced with the 6 categories determined by the bank side. The corresponding relationship of the inner item categories is predefined by the bank side, including the mapping relationship between the inner item categories stipulated by the express company and the inner item categories stipulated by the bank side.

[0103] In an embodiment, as Figure 4 shown, the steps of creating the risk warning model include:

[0104] Step S201, obtaining a plurality of sample dynamic data authorized by the user himself / herself, where the dynamic data includes sample work and rest data and sample express delivery data.

[0105] Specifically, the dynamic data obtained in this step is that of multiple different users. The sample dynamic data includes sample work and rest data and sample express delivery data. Therefore, the sample work and rest data and sample express delivery data here are obtained in a similar manner to the dynamic data of the user who submits the service handling request in step S102. To ensure the accuracy of the training model, it is necessary to obtain the sample dynamic data of a large number of users here. The sample dynamic data of each user is obtained under the authorization of the user himself and a third-party institution (such as the express delivery company that provides express delivery data), and is only used in the process of training the risk warning model in this application and will not be used for other purposes.

[0106] Step S202, generate a training data set according to the multiple sample dynamic data.

[0107] The training data set includes the sample dynamic data of a large number of different users, that is, the relationship between the user and the dynamic data is one-to-one. The dynamic data corresponding to each user also includes work and rest data and express delivery data within a preset time range. In the training data set, it also includes the classification label corresponding to each customer. Different classification labels represent different risk assessment levels. The method for determining the classification label will be described in detail in the subsequent embodiments.

[0108] Step S203, use the training data set to train a pre-established machine learning model to obtain the risk warning model.

[0109] Specifically, input each data in the training data set into the pre-built machine learning model, so that the machine learning model learns and classifies the data. The trained risk warning model can obtain the risk assessment result of the user according to the input dynamic data of the user to be evaluated.

[0110] During training, it is usually necessary to perform multiple rounds of training and verification on the risk warning model to achieve the overdue effect. For example, the method for model verification is to obtain the credit evaluation situation of the users corresponding to each sample dynamic data through channels such as the credit investigation system, and judge whether the classification result of the risk warning model for the sample dynamic data is consistent with the credit evaluation situation of the corresponding user. If they are consistent, it means that the reliability degree of the risk warning model is relatively high. Otherwise, the risk warning model needs to be further trained. The consistency between the classification result of the sample dynamic data and the credit evaluation situation of the corresponding user here refers to the consistency in scope. For example, if the user's credit evaluation is good, and the classification result of the risk warning model for the user is low risk, it can be considered that the classification result of the risk warning model for the sample dynamic data is consistent with the credit evaluation situation of the corresponding user.

[0111] In one embodiment, as Figure 5 shown, step S202, generate a training data set according to the multiple sample dynamic data, includes steps S2021 to S2023:

[0112] In step S2021, data cleaning and data governance are performed on the multiple sample dynamic data to obtain initial data.

[0113] Specifically, the process of data cleaning and data governance for the multiple sample dynamic data is similar to the process of processing the dynamic data of the user to be evaluated in step S103, and reference can be made to the implementation.

[0114] In step S2022, classification labels for the initial data are determined according to preset label rules.

[0115] Here, it is necessary to perform a preliminary analysis on the data in the initial data according to the preset label rules to obtain the classification label results corresponding to each initial data. Specifically, as Figure 6 shown, step S2022 is further implemented through the following steps:

[0116] In step S20221, the sending time, receiving time, working time, and off - work time of the same user in the initial data are obtained. This step summarizes the data of the user's sending and receiving time and working time.

[0117] In step S20222, the high - frequency sending and receiving time periods of the user are determined according to the sending time and the receiving time.

[0118] In this step, by analyzing the user's sending and receiving time, the corresponding relationship between the sending and receiving time and the date is established, and then the user's high - frequency sending and receiving time periods and the change trend of the high - frequency sending and receiving time periods are obtained.

[0119] In step S20223, the working time period of the user is determined according to the working time and the off - work time.

[0120] In this step, by analyzing the user's working and off - work time, the corresponding relationship between the working and off - work time and the date is established, and then the user's working time period and non - working time period, as well as the change trend of the working time period and non - working time period, are obtained.

[0121] In step S20224, the classification labels are determined according to the high - frequency sending and receiving time periods and the working time period.

[0122] Specifically, (1) a first score is determined according to the matching degree between the high - frequency sending and receiving time periods and the working time period and a preset scoring table.

[0123] It is considered that the business behavior of the user is relatively healthy if there are stable pick-up and delivery time periods in the user's express delivery behavior and the pick-up and delivery time periods basically coincide with the working time periods. The matching degree between the high-frequency pick-up and delivery time periods and the working time periods in this step refers to the coincidence degree between the pick-up and delivery time periods and the working time periods, and the high-frequency pick-up and delivery time periods and the working time periods within the same date cycle should be matched. If the coincidence degree between the high-frequency pick-up and delivery time periods and the working time periods of the user within the same date cycle reaches the preset threshold, it is considered that the user conducts express delivery pick-up and delivery during normal working and living hours. At this time, the corresponding score can be found according to the preset first score table. The first score table includes the mapping relationship between the coincidence degree between the pick-up and delivery time periods and the working time periods and the first score.

[0124] (2) Determine the second score according to the stability degree of the high-frequency pick-up and delivery time periods and the score table.

[0125] Here, it is considered whether the business behavior of the user is stable. The stability here does not only refer to the instability of the user's high-frequency pick-up and delivery time periods, but also includes the situation where the changes in the user's high-frequency pick-up and delivery time periods are consistent with the working time periods. The second score can also be determined by simply looking it up according to the preset second score table. The second score table includes the mapping relationship between the change rule of the pick-up and delivery time periods and the working time periods and the second score.

[0126] (3) Determine the third score according to the matching degree between the high-frequency pick-up and delivery time periods and the preset abnormal time periods and the score table.

[0127] Here, it is considered whether the business behavior of the user is abnormal. The preset abnormal time periods vary from user to user and are specifically determined according to the user's working time periods. For example, if the user's working time is during the day, then 2 am to 3 am belongs to the abnormal time period; if the user's working time is at night (the user works the night shift), then 2 am to 3 am is within their working time period and does not belong to the abnormal time period. If the coincidence degree between the user's high-frequency pick-up and delivery time periods and the corresponding abnormal time periods is higher than the preset threshold, it is considered that the business behavior of this customer is abnormal. At this time, the corresponding score can be found according to the preset third score table. The third score table includes the mapping relationship between the coincidence degree between the pick-up and delivery time periods and the abnormal time periods and the third score.

[0128] (4) Determine the classification label according to the sum of the first score, the second score, and the third score and the mapping relationship between the preset classification label and the score.

[0129] Specifically, the values of the first score, the second score, and the third score can be positive or negative. By performing a summation operation, the sum of the three score values can be obtained. At this time, based on the preset mapping relationship between the classification labels and the scores, the classification label can be determined. The classification labels can be, for example, 0, 1, 2, and 3, representing extremely high risk, high risk, relatively low risk, and low risk respectively. In practical applications, different granularities can be set according to requirements, and the corresponding relationship between the high and low scores and the high and low risk levels is also determined according to the set rules. This application only provides an example and is not a limitation to this application.

[0130] Step S2023: Generate the training dataset according to the initial data and its corresponding classification label.

[0131] This embodiment only takes three scores as an example for illustration. In practical applications, more or fewer perspectives can be used as the basis for determining the classification label.

[0132] This application conducts user risk assessment based on the user's dynamic data, and more specifically, combines the user's express delivery data and work and rest data for assessment, solving the shortcoming problem of the existing dynamic assessment method for customer risk warning in banks, and providing a more dynamic, vivid, real-time, high-value, and brand-new method for user assessment in the process of bank business handling.

[0133] Based on the same inventive concept, the embodiment of this application also provides a bank user risk warning device based on big data, which can be used to implement the method described in the above embodiment, as described in the following embodiment. Since the principle of the bank user risk warning device based on big data to solve problems is similar to that of the bank user risk warning method based on big data, the implementation of the bank user risk warning device based on big data can refer to the implementation of the bank user risk warning method based on big data, and the repeated parts will not be elaborated. Hereinafter, the term "unit" or "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the systems described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0134] As Figure 7 shown, this application also provides a bank user risk warning device based on big data, including:

[0135] A request receiving module 301, configured to receive a business handling request submitted by a user;

[0136] A dynamic data acquisition module 302, configured to acquire the dynamic data of the user authorized by the user and a third party institution according to the business handling request;

[0137] A data preprocessing module 303, configured to perform data cleaning and data governance on the dynamic data to obtain the evaluation data of the user;

[0138] A risk assessment module 304, configured to determine the risk assessment result of the user according to the evaluation data and a pre-created risk warning model; wherein, the risk warning model is pre-trained according to the dynamic data of multiple users and the corresponding ones.

[0139] In one embodiment, as Figure 8 shown, the dynamic data acquisition module 302 includes:

[0140] A work and rest data acquisition unit 3021, configured to acquire the work and rest data of the user; the work and rest data includes the user's work start time, work end time, work address, and residential address; the work and rest data is acquired under the authorization of the user.

[0141] A courier data acquisition unit 3022, configured to acquire the courier data of the user from multiple courier companies; the courier data includes the courier company brand, recipient address, sender address, recipient information, sender information, internal item category, sending time, receiving time, and receiving method; the courier data is acquired under the authorization of the user and the courier company.

[0142] In one embodiment, the data preprocessing module 303 is specifically configured to:

[0143] Screen the dynamic data according to a preset evaluation time range; and

[0144] Convert the screened dynamic data into the evaluation data according to a predefined data format.

[0145] In one embodiment, as Figure 9 shown, the bank user risk warning device based on big data further includes:

[0146] A sample data acquisition module 305, configured to acquire multiple sample dynamic data authorized by the user himself, and the dynamic data includes sample work and rest data and sample courier data;

[0147] A training data generation module 306, configured to generate a training data set according to the multiple sample dynamic data;

[0148] A model training module 307, configured to use the training data set to train a pre-established machine learning model to obtain the risk warning model.

[0149] In one embodiment, as Figure 10 shown, the training data generation module 306 includes:

[0150] A data preprocessing unit 3061, configured to perform data cleaning and data governance on the multiple sample dynamic data to obtain initial data;

[0151] A classification label determination unit 3062, configured to determine the classification label of the initial data according to a preset label rule;

[0152] A training data generation unit 3063, configured to generate the training data set according to the initial data and its corresponding classification label.

[0153] In one embodiment, as Figure 11 shown, the classification label determination unit 3062 includes:

[0154] A concerned data acquisition subunit 30621, configured to acquire the sender time, recipient time, work start time, and work end time of the same user in the initial data;

[0155] A high-frequency sending and receiving period acquisition subunit 30622, configured to determine the high-frequency sending and receiving period of the user according to the sender time and the recipient time;

[0156] A work period acquisition subunit 30623, configured to determine the work period of the user according to the work start time and the work end time;

[0157] A classification label determination subunit 30624, configured to determine the classification label according to the high-frequency sending and receiving period and the work period.

[0158] In one embodiment, the classification label determination subunit 30624 is specifically configured to:

[0159] Determine a first score according to the matching degree between the high-frequency sending and receiving period and the work period and a preset scoring table;

[0160] Determine a second score according to the stability degree of the high-frequency sending and receiving period and the scoring table;

[0161] Determine a third score according to the matching degree between the high-frequency sending and receiving period and a preset abnormal period and the scoring table;

[0162] Determine the classification label according to the sum of the first score, the second score, and the third score and a preset mapping relationship between the classification label and the score.

[0163] This application performs user risk assessment based on the dynamic data of users. More specifically, it combines the express delivery data and work and rest data of users for assessment, solves the shortcoming problem of the existing dynamic assessment method for customer risk warning in banks, and provides a more dynamic, vivid, real-time, high-value, and brand-new method for user assessment in the process of banking business handling.

[0164] In one embodiment, the present application further provides a computer device. Refer to Figure 12 , the electronic device 100 specifically includes:

[0165] A central processing unit (processor) 110, a memory 120, a communication module (Communications) 130, an input unit 140, an output unit 150, and a power supply 160.

[0166] Among them, the memory 120, the communication module (Communications) 130, the input unit 140, the output unit 150, and the power supply 160 are respectively connected to the central processing unit (processor) 110. A computer program is stored in the memory 120, the central processing unit 110 can call the computer program, and when the central processing unit 110 executes the computer program, all steps in the method for early warning of bank user risks based on big data in the above embodiments are implemented.

[0167] In one embodiment, an embodiment of the present application further provides a computer-readable storage medium for storing a computer program, and the computer program can be executed by a processor. When the computer program is executed by the processor, any method for early warning of bank user risks provided by the present invention is implemented.

[0168] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0169] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0170] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to work in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction device that implements the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 specified in the block or multiple blocks.

[0171] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 specified in the block or multiple blocks.

[0172] The specific embodiments described above further elaborate on the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for early warning of bank user risks based on big data, characterized in that, It includes: Receiving a service handling request submitted by a user; Obtaining the dynamic data of the user authorized by the user and a third-party institution according to the service handling request, where the dynamic data includes work and rest data and express delivery data; Performing data cleaning and data governance on the dynamic data to obtain the evaluation data of the user; Determining the risk assessment result of the user according to the evaluation data and a pre-created risk warning model; wherein, the risk warning model is pre-trained according to the dynamic data of multiple users; Wherein, the dynamic data is the data provided by the third-party institution after data decryption and encrypted transmission; Wherein, the steps of creating the risk warning model include: Obtaining multiple sample dynamic data authorized by the user himself, where the dynamic data includes sample work and rest data and sample express delivery data; Generating a training data set according to the multiple sample dynamic data; Using the training data set to train a pre-established machine learning model to obtain the risk warning model; Wherein, generating the training data set according to the multiple sample dynamic data includes: Performing data cleaning and data governance on the multiple sample dynamic data to obtain initial data; determining the classification labels of the initial data according to preset label rules; Generating the training data set according to the initial data and its corresponding classification labels; Wherein, determining the classification labels of the initial data according to the preset label rules includes: obtaining the sending time, receiving time, working time, and off-work time of the same user in the initial data; Determining the high-frequency receiving and sending time periods of the user according to the sending time and the receiving time; Determining the working time period of the user according to the working time and the off-work time; Determining the classification labels according to the high-frequency receiving and sending time periods and the working time period; Wherein, determining the classification labels according to the high-frequency receiving and sending time periods and the working time period includes: Determining the first score according to the matching degree between the high-frequency receiving and sending time periods and the working time period and a preset scoring table; Determining the second score according to the stability degree of the high-frequency receiving and sending time periods and the scoring table; determining the third score according to the matching degree between the high-frequency receiving and sending time periods and a preset abnormal time period and the scoring table; Determining the classification labels according to the sum of the first score, the second score, and the third score and the mapping relationship between the preset classification labels and the scores.

2. The method for warning of bank user risks based on big data according to claim 1, wherein The obtaining the dynamic data of the user authorized by the user and a third-party institution according to the service handling request includes: Obtaining the work and rest data of the user; the work and rest data includes the user's working time, off-work time, work address, and residential address; the work and rest data is obtained under the authorization of the user; Obtaining the express delivery data of the user from multiple express delivery companies; the express delivery data includes the express delivery company brand, receiving address, sending address, recipient information, sender information, internal item category, sending time, receiving time, and receiving method; the express delivery data is obtained under the authorization of the user and the express delivery company.

3. The method for warning of bank user risks based on big data according to claim 1, wherein The performing data cleaning and data governance on the dynamic data to obtain the evaluation data of the user includes: Filter the dynamic data according to a preset evaluation time range; and Convert the filtered dynamic data into the evaluation data according to a predefined data format.

4. A bank user risk early warning device based on big data, characterized in that, It includes: A request receiving module, configured to receive a service handling request submitted by a user; A dynamic data acquisition module, configured to acquire the dynamic data of the user authorized by the user and a third-party institution according to the service handling request, where the dynamic data includes work and rest data and express delivery data; A data preprocessing module, configured to perform data cleaning and data governance on the dynamic data to obtain the evaluation data of the user; A risk assessment module, configured to determine the risk assessment result of the user according to the evaluation data and a pre-created risk warning model; where the risk warning model is pre-trained according to the dynamic data of multiple users; Wherein, the dynamic data is the data provided by the third-party institution after data decryption and encrypted transmission; Wherein, the steps of creating the risk warning model include: Acquire multiple sample dynamic data authorized by the user himself, where the dynamic data includes sample work and rest data and sample express delivery data; Generate a training data set according to the multiple sample dynamic data; Use the training data set to train a pre-established machine learning model to obtain the risk warning model; Wherein, generating the training data set according to the multiple sample dynamic data includes: Perform data cleaning and data governance on the multiple sample dynamic data to obtain initial data; determine the classification labels of the initial data according to preset label rules; Generate the training data set according to the initial data and its corresponding classification labels; Wherein, determining the classification labels of the initial data according to the preset label rules includes: acquiring the sending time, receiving time, working time, and off-work time of the same user in the initial data; Determine the high-frequency sending and receiving period of the user according to the sending time and the receiving time; Determine the working period of the user according to the working time and the off-work time; Determine the classification labels according to the high-frequency sending and receiving period and the working period; Wherein, determining the classification labels according to the high-frequency sending and receiving period and the working period includes: Determine a first score according to the matching degree between the high-frequency sending and receiving period and the working period and a preset scoring table; Determine a second score according to the stability degree of the high-frequency sending and receiving period and the scoring table; determine a third score according to the matching degree between the high-frequency sending and receiving period and a preset abnormal period and the scoring table; Determine the classification labels according to the sum of the first score, the second score, and the third score and the mapping relationship between the preset classification labels and the scores.

5. A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the big data-based bank user risk warning method according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the big data-based bank user risk warning method according to any one of claims 1 to 3.

Images