Risk early warning method, system and equipment based on adaptive algorithm selection and medium

Abstract

The invention discloses a risk early warning method, system and device based on adaptive algorithm selection and a medium, and relates to the technical field of risk early warning, and the method comprises the steps: extracting historical integral data, constructing an original data set, extracting basic parameters, constructing comprehensive risk characteristic parameters, and constructing a characteristic space description matrix; calculating complexity indexes of algorithms in the algorithm library and fitness indexes on the feature space description matrix, constructing an optimization objective function, solving the optimization objective function, selecting an optimal algorithm, performing training, generating a risk control model, dividing users into different groups according to output of the risk control model, generating attribute tags, constructing a risk control rule set, and performing risk control according to the risk control rule set. And performing risk assessment on each new transaction, calculating a risk score, and performing risk early warning. Through basic parameters, comprehensive risk feature parameters and a multi-objective optimization method, algorithm complexity and adaptation are comprehensively considered, and high adaptability, comprehensive feature extraction and optimization of algorithm complexity and adaptation of risk early warning are realized.

Description

Technical Field

[0001] The present invention relates to the field of risk warning technology, and in particular to a risk warning method, system, equipment and medium based on adaptive algorithm selection. Background Art

[0002] With the widespread use of e-commerce and point systems, it has become normal for users to conduct various transactions through points. Such systems encourage users to participate in transactions frequently through a point reward mechanism, thereby improving the user stickiness and activity of the platform. However, the openness and convenience of the point system also bring potential risks.

[0003] Existing risk management systems usually use static rule engines or single machine learning algorithms for risk assessment. These methods have the following main problems: insufficient adaptability. Static rules are difficult to cope with the ever-changing user behavior patterns and new fraud methods, and cannot be dynamically adjusted to adapt to new risk characteristics; limited feature utilization. Traditional methods are relatively simple in feature extraction and utilization, and cannot fully mine the multi-dimensional information in user transaction data, resulting in insufficient accuracy and comprehensiveness in risk identification; algorithm selection is not optimized. The performance of a single algorithm varies greatly under different data features and business scenarios. There is a lack of adaptive selection mechanism, and it is impossible to ensure that the optimal algorithm can be selected in various situations, affecting the overall risk warning effect; complexity and performance conflict. While pursuing high accuracy, existing systems often ignore the computational complexity of the algorithm, resulting in obvious performance bottlenecks when the system processes large-scale data, affecting real-time warning capabilities.

[0004] In response to the above problems, there is an urgent need for a risk warning method that can dynamically adapt to data changes, autonomously select the optimal algorithm, and make full use of multi-dimensional risk characteristics to improve the security and operational efficiency of the points system. Summary of the invention

[0005] Based on the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a risk warning method, system, device and medium based on adaptive algorithm selection to solve the above-mentioned technical problems.

[0006] To achieve the above object, the present invention provides the following technical solution: a risk early warning method based on adaptive algorithm selection, comprising: S1: extracting historical integral data from the integral system database, constructing an original data set, and preprocessing the original data set; S2: extracting basic parameters from the preprocessed original data set, and constructing comprehensive risk characteristic parameters based on the basic parameters; S3: Define an algorithm library containing multiple machine learning algorithms, construct a feature space description matrix based on basic parameters and comprehensive risk feature parameters, calculate the complexity index of each algorithm in the algorithm library, and calculate the fitness index of the algorithm on the feature space description matrix; S4: Construct an optimization objective function based on the complexity index and the fitness index, select the optimal algorithm by solving the optimization objective function, use the selected algorithm to train the feature matrix, generate a risk control model, and divide users into different groups and generate attribute labels based on the output of the risk control model; S5: Based on the current attribute labels, a risk control rule set is constructed to conduct risk assessment on each new transaction and calculate the risk score. When the risk score is less than the preset threshold, a risk warning is issued and a risk warning report is generated.

[0007] The present invention is further configured such that the basic parameters include user information and transaction information, and the transaction information includes regional code, transaction time interval, average transaction time interval, number of transaction location changes, total number of device types and frequency of user device use, points change amount, points change type and points acquisition and use ratio.

[0008] The present invention is further configured such that the comprehensive risk characteristic parameters include spatiotemporal volatility characteristics, transaction pattern diversity characteristics, and point utilization characteristics; The spatiotemporal volatility characteristics are calculated based on the trading time interval, the average trading time interval, and the number of trading location changes; Calculate transaction pattern diversity characteristics based on the total number of device types and user device usage frequency; The integral utilization characteristics are calculated based on the integral change amount, integral change type, and integral acquisition and utilization ratio.

[0009] The present invention is further configured such that the calculation logic of the spatiotemporal volatility feature is: ,in, For users The spatiotemporal volatility characteristics of For users In the The time interval between the previous transaction and the next transaction. For users The average transaction time interval, For users The total number of transactions, is the number of transaction location changes, is the maximum number of transaction location changes, is the power exponent parameter, To adjust the parameters; The calculation logic of the transaction pattern diversity feature is: ,in, For users The diversity of transaction patterns, is the total number of device types, For users No. The frequency of user device usage, and is the power exponent parameter, is the total number of time periods in a day, For the The proportion of transactions conducted in a time period; The calculation logic of the points utilization feature is: ,in, For users The points utilization feature, For users In the The change in points for each transaction, is the type of integral change, coded as a categorical variable, The ratio of points earned and used. is the power exponent parameter, To adjust the parameters.

[0010] The present invention is further configured to construct a feature space description matrix according to the basic parameters and the comprehensive risk feature parameters, including: Construct the initial feature vector by using the basic parameters and comprehensive risk characteristic parameters in any order ,in, , is the number of basic parameters and comprehensive risk characteristic parameters, For the basic parameters or comprehensive risk characteristic parameters; therefore, the initial feature space description matrix ,and , is the number of users, is a mathematical symbol representing the set of real numbers; The initial feature space description matrix Convert to polynomial feature space description matrix , , where the basis function Defined as: , is the total number of polynomial features, is the initial eigenvector Middle Features, is a polynomial basis function For The power of features, and satisfies ; The initial feature space description matrix and the polynomial feature space description matrix splicing, and generate the final feature space description matrix through principal component analysis and dimensionality reduction, where , is the final feature space description matrix, For principal component analysis, is the feature weight matrix after dimensionality reduction.

[0011] The present invention is further configured such that the calculation logic of the complexity index of each algorithm in the algorithm library is: ,in, For the The complexity index of an algorithm is is the number of feature dimensions of the feature weight matrix after dimensionality reduction, For the The importance weight of the feature, For the The algorithm for The sensitivity of a feature, No. The discreteness of the features, To adjust the parameters; The calculation logic of the algorithm's fitness index on the feature space description matrix is: according to the final feature space description matrix Constructing the kernel matrix , ,in, is a similarity evaluation function used to evaluate the similarity between users. ,in, For the The fitness index of an algorithm is For users and users The similarity evaluation function value of For the The algorithm is used by the user and users The similarity evaluation value of .

[0012] The present invention is further configured that, in step S4, the optimization objective function is: ,in, To optimize the objective function, and is the weight coefficient, and the sum is 1; Select the algorithm that optimizes the minimum value of the objective function , according to the algorithm Train the feature matrix to generate a risk control model. Based on the output of the risk control model, divide users into different groups and generate attribute labels. In step S5, the risk control rule set is a set of multiple parameter judgments, wherein each parameter is set with a preset value or interval; Get the attribute label of the user of each new transaction, get the risk control rule set of the attribute label, judge the parameters of the new transaction, when the conditions are met, the score value of the parameter is set to 1, when the conditions are not met, the score value of the parameter is set to 0, and the weighted sum of all parameters of the new transaction is obtained to obtain the risk score.

[0013] The present invention also provides a risk warning system based on adaptive algorithm selection, which is used to implement the above-mentioned risk warning method based on adaptive algorithm selection, and the system includes: Data extraction module: extracting historical integral data from the integral system database, constructing an original data set, and preprocessing the original data set; Feature construction module: extracting basic parameters from the preprocessed original data set and constructing comprehensive risk feature parameters based on the basic parameters; Calculation module: defines an algorithm library containing multiple machine learning algorithms, constructs a feature space description matrix based on basic parameters and comprehensive risk feature parameters, calculates the complexity index of each algorithm in the algorithm library, and calculates the fitness index of the algorithm on the feature space description matrix; Division module: Construct an optimization objective function based on the complexity index and the fitness index, select the optimal algorithm by solving the optimization objective function, use the selected algorithm to train the feature matrix, generate a risk control model, and divide users into different groups and generate attribute labels based on the output of the risk control model; Risk warning module: Based on the current attribute tags, a risk control rule set is constructed to conduct risk assessment on each new transaction and calculate the risk score. When the risk score is less than the preset threshold, a risk warning is issued and a risk warning report is generated.

[0014] The present invention further provides an electronic device, comprising: one or more processors; A storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the electronic device implements the risk warning method based on adaptive algorithm selection as described in any one of the above.

[0015] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor of a computer, the computer is enabled to execute the risk warning method based on adaptive algorithm selection as described in any one of the above items.

[0016] The present invention provides a risk warning method, system, device and medium based on adaptive algorithm selection. The method extracts historical integral data from a points system database, constructs an original data set, and preprocesses the original data set; extracts basic parameters from the preprocessed original data set, and constructs comprehensive risk feature parameters based on the basic parameters; defines an algorithm library containing multiple machine learning algorithms, constructs a feature space description matrix according to the basic parameters and the comprehensive risk feature parameters, calculates the complexity index of each algorithm in the algorithm library, and calculates the fitness index of the algorithm on the feature space description matrix; constructs an optimization objective function according to the complexity index and the fitness index, selects the optimal algorithm by solving the optimization objective function, trains the feature matrix using the selected algorithm, generates a risk control model, divides users into different groups and generates attribute labels according to the output of the risk control model; constructs a risk control rule set based on the current attribute label, performs risk assessment on each new transaction, calculates the risk score, and when the risk score is less than a preset threshold, performs risk warning and generates a risk warning report, and the beneficial effects produced include: 1. Strong adaptability: By defining an algorithm library containing a variety of machine learning algorithms, and dynamically selecting the optimal algorithm based on the fitness index and complexity index on the feature space description matrix. This adaptive algorithm selection mechanism can adjust the algorithm selection in real time according to changes in data characteristics and business needs, significantly improving the flexibility and adaptability of the risk warning system, and overcoming the problem of static algorithm selection and lack of adaptability in existing technologies; 2. Comprehensive feature extraction: By extracting basic parameters and constructing comprehensive risk feature parameters, including spatiotemporal volatility features, transaction pattern diversity features, and point utilization features, it fully reflects the multi-dimensional behavior patterns of users in the points system. This multi-dimensional feature extraction method can more comprehensively capture the complexity and potential risks in user behavior, and improve the accuracy and comprehensiveness of risk identification; 3. Algorithm complexity and adaptability optimization: In the algorithm selection process, the algorithm complexity index and adaptability index are comprehensively considered, and the algorithm that achieves the best balance between complexity and adaptability is selected through multi-objective optimization methods. This not only ensures the accuracy of risk warning, but also takes into account the system's computing efficiency and real-time performance, and improves the performance of the overall risk control system.

[0017] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 A flow chart of a risk warning method based on adaptive algorithm selection is shown as an exemplary embodiment of the present invention; Figure 2 The schematic diagram of the structure of a risk early warning system based on adaptive algorithm selection is shown as an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0019] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.

[0020] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0021] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0022] Embodiment 1 Risk early warning methods based on adaptive algorithm selection, such as Figure 1 As shown, including: S1: extracting historical integral data from the integral system database, constructing an original data set, and preprocessing the original data set; S2: extracting basic parameters from the preprocessed original data set, and constructing comprehensive risk characteristic parameters based on the basic parameters; S3: Define an algorithm library containing multiple machine learning algorithms, construct a feature space description matrix based on basic parameters and comprehensive risk feature parameters, calculate the complexity index of each algorithm in the algorithm library, and calculate the fitness index of the algorithm on the feature space description matrix; S4: Construct an optimization objective function based on the complexity index and the fitness index, select the optimal algorithm by solving the optimization objective function, use the selected algorithm to train the feature matrix, generate a risk control model, and divide users into different groups and generate attribute labels based on the output of the risk control model; S5: Based on the current attribute labels, a risk control rule set is constructed to conduct risk assessment on each new transaction and calculate the risk score. When the risk score is less than the preset threshold, a risk warning is issued and a risk warning report is generated.

[0023] The present invention is further configured such that the basic parameters include user information and transaction information, and the transaction information includes region code, transaction time interval, average transaction time interval, number of transaction location changes, total number of device types and frequency of use of user devices, amount of points change, type of points change and ratio of points acquisition and use. Specifically, user information refers to basic information and characteristic data related to each user in the system. This information is used to identify the identity of the user, analyze the user's behavior pattern and evaluate the user's risk level; transaction information contains all transaction activity records of the user in the points system, which is the core data for evaluating user behavior and potential risks. In the transaction information, the region code indicates the geographical location of the user when the transaction is conducted, which is expressed in the form of a code of a city or a specific location, and is used to analyze the user's geographical distribution and behavior pattern; the transaction time interval refers to the time difference between two consecutive transactions of the user, reflecting the frequency and rhythm of the user's transactions; the average transaction time interval is the average of all transaction time intervals of the user within a specific time window, reflecting the overall transaction frequency of the user; the number of transaction location changes refers to the number of times the transaction location of the user changes within a specific time window, which is used to evaluate the user's geographical behavior diversity and potential risk activities; the device type The total number of types refers to the number of different types of devices used by users when conducting transactions, reflecting the diversity of users' device usage; the frequency of user device usage refers to the number of times a user uses each type of device to conduct transactions within a specific time window, reflecting the user's dependence on different devices and usage habits; the amount of point change refers to the increase or decrease in the number of points of a user in each transaction, reflecting the user's point acquisition and consumption behavior; the point change type refers to the category of point increase or decrease. In a feasible embodiment of the present invention, point acquisition is recorded as 1 and point use is recorded as 1.2; the point acquisition and use ratio refers to the ratio of points acquired by a user to points used within a certain time window, reflecting the balance of the user's point acquisition and consumption behavior; preprocessing includes data cleaning, missing value processing and smoothing processing, which is a prior art and will not be explained here; The present invention is further configured such that the comprehensive risk characteristic parameters include spatiotemporal volatility characteristics, transaction pattern diversity characteristics, and point utilization characteristics; The spatiotemporal volatility feature is calculated according to the transaction time interval, the average transaction time interval and the number of transaction location changes; the present invention is further configured that the calculation logic of the spatiotemporal volatility feature is: ,in, For users The spatiotemporal volatility characteristics of For users In the The time interval between the previous transaction and the next transaction. For users The average transaction time interval, For users The total number of transactions, is the number of transaction location changes, is the maximum number of transaction location changes, is the power exponent parameter, is an adjustment parameter; specifically, the spatiotemporal volatility feature is used to measure the degree of change in the time and location of the user's trading behavior in the points system. By analyzing the user's trading time interval, average trading time interval, and the number of changes in trading locations, the spatiotemporal volatility feature can reflect the stability and abnormality of the user's trading behavior, thereby assisting the risk warning system to identify potential high-risk transactions; the power index parameter is a positive real number, ranging from [1,2], and the adjustment parameter is also a positive real number with a value range of (0,1]. By comprehensively analyzing the transaction time interval and the number of location changes, the spatiotemporal volatility feature can accurately reflect the volatility of user trading behavior and help identify abnormal trading patterns. Combining the fluctuations in the two dimensions of time and location, the spatiotemporal volatility feature provides a more comprehensive user behavior feature, significantly improving the risk warning system's ability to identify potential risks. The transaction mode diversity feature is calculated based on the total number of device types and the frequency of use of user devices; the calculation logic of the transaction mode diversity feature is: ,in, For users The diversity of transaction patterns, is the total number of device types, For users No. The frequency of user device usage, and is the power exponent parameter, is the total number of time periods in a day, For the The transaction pattern diversity feature is designed to measure the diversity of transactions made by users using different types of devices in the points system. By combining the device usage frequency and transaction time distribution, it comprehensively reflects the diversity of users' device usage habits and transaction behaviors. and All are positive real numbers with a value range of [1,2]. The transaction pattern diversity feature combines the two key dimensions of device usage frequency and transaction time distribution, provides a comprehensive assessment of user transaction patterns, and can more accurately reflect the complexity and diversity of user behavior. By capturing user transaction behaviors on different devices and time periods, the transaction pattern diversity feature helps to identify abnormal or high-risk transaction patterns and improve the accuracy and effectiveness of the risk warning system. The point utilization feature is calculated according to the point change amount, point change type and point acquisition and use ratio; the calculation logic of the point utilization feature is: ,in, For users The points utilization feature, For users In the The change in points for each transaction, is the type of integral change, coded as a categorical variable, The ratio of points earned and used is: is the power exponent parameter, To adjust the parameters, specifically, the points utilization feature aims to measure the efficiency and balance of users' acquisition and use of points in the points system. By comprehensively considering the amount of points change, the type of points change, and the proportion of points acquisition and use in each transaction, the points utilization feature can reflect the activity and rationality of users' points activities, thereby assisting the risk warning system to identify abnormal or high-risk points operation behaviors; power index parameter is a positive real number, ranging from [1,2], and the adjustment parameter is also a positive real number with a value range of (0,1]. The point utilization feature comprehensively considers the amount and type of user point acquisition and use, can accurately reflect the balance and efficiency of user point activities, and assist in identifying abnormal point operation behaviors. By combining the amount, type, and acquisition and use ratio of points, the point utilization feature provides multi-dimensional risk characteristics, which significantly improves the risk warning system's ability to identify potential risks.

[0024] The present invention is further configured to construct a feature space description matrix according to the basic parameters and the comprehensive risk feature parameters, including: Construct the initial feature vector by using the basic parameters and comprehensive risk characteristic parameters in any order ,in, , is the number of basic parameters and comprehensive risk characteristic parameters, For the basic parameters or comprehensive risk characteristic parameters; therefore, the initial feature space description matrix ,and , is the number of users, is a mathematical symbol representing the set of real numbers; The initial feature space description matrix Convert to polynomial feature space description matrix , , where the basis function Defined as: , is the total number of polynomial features, is the initial eigenvector Middle Features, is a polynomial basis function For The power of features, and satisfies ; The initial feature space description matrix and the polynomial feature space description matrix splicing, and generate the final feature space description matrix through principal component analysis and dimensionality reduction, where , is the final feature space description matrix, For principal component analysis, is the feature weight matrix after dimensionality reduction.

[0025] The present invention is further configured such that the calculation logic of the complexity index of each algorithm in the algorithm library is: ,in, For the The complexity index of an algorithm is is the number of feature dimensions of the feature weight matrix after dimensionality reduction, For the The importance weight of the feature, For the The algorithm for The sensitivity of a feature, For the The discreteness of the features, is the adjustment parameter; specifically, The algorithm for The sensitivity of a feature measures the The degree of dependence on the feature, that is, The influence of the change of each feature on the output of the algorithm is The method of perturbing each feature with a small amplitude and observing the change of the algorithm output to quantify the sensitivity is a prior art and will not be described in detail here. The discreteness of a feature measures the distribution breadth and diversity of the feature value in the data set, reflecting the variability and information content of the feature. The information entropy is used to measure the discreteness of the feature. The higher the information entropy, the more uniform the feature distribution and the greater the discreteness. This is a prior art and will not be elaborated here. The calculation logic of the algorithm's fitness index on the feature space description matrix is: according to the final feature space description matrix Constructing the kernel matrix , ,in, is a similarity evaluation function used to evaluate the similarity between users. ,in, For the The fitness index of an algorithm is For users and users The similarity evaluation function value of For the The algorithm is used by the user and users The similarity evaluation value of .

[0026] The present invention is further configured that, in step S4, the optimization objective function is: ,in, To optimize the objective function, and is the weight coefficient, and the sum is 1; Select the algorithm that optimizes the minimum value of the objective function , according to the algorithm The feature matrix is ​​trained to generate a risk control model, and users are divided into different groups and attribute labels are generated according to the output of the risk control model. Specifically, multiple machine learning algorithms include classification algorithms, clustering algorithms, and deep learning algorithms. The feature matrix is ​​trained to generate a risk control model, and users are divided into different groups and attribute labels are generated according to the output of the risk control model. All of these can be achieved through the above-mentioned machine learning algorithms, which is a prior art and is not limited here.

[0027] In step S5, the risk control rule set is a set of multiple parameter judgments, wherein each parameter is set with a preset value or interval; Obtain the attribute label of the user of each new transaction, obtain the risk control rule set of the attribute label, judge the parameters of the new transaction, and set the score value of the parameter to 1 when the condition is met; when the condition is not met, the score value of the parameter is set to 0. Perform weighted summation on all parameters of the new transaction to obtain the risk score. Specifically, according to the user ID of the new transaction, obtain the attribute label of the user from the existing risk control model. The attribute label reflects the user's behavior pattern, risk preference and other characteristics. According to the user's attribute label, extract the risk control rule set applicable to the user group from the predefined risk control rule library. These rules are designed for different user characteristics and cover multi-dimensional risk control parameters. Apply the risk control rules to each parameter of the new transaction one by one for judgment. If the transaction parameter meets the condition of a certain risk control rule, the score value of the parameter is set to 1; if not, it is set to 0. Perform weighted summation on the score values ​​of all parameters according to the preset weights to obtain the final risk score of the new transaction. According to the comparison result between the risk score and the preset threshold, decide whether to trigger the risk warning and generate the corresponding risk warning report.

[0028] Embodiment 2 See also Figure 2 The exemplary risk warning system based on adaptive algorithm selection is used to implement the above-mentioned risk warning method based on adaptive algorithm selection, and the system includes: Data extraction module: extracting historical integral data from the integral system database, constructing an original data set, and preprocessing the original data set; Feature construction module: extracting basic parameters from the preprocessed original data set and constructing comprehensive risk feature parameters based on the basic parameters; Calculation module: defines an algorithm library containing multiple machine learning algorithms, constructs a feature space description matrix based on basic parameters and comprehensive risk feature parameters, calculates the complexity index of each algorithm in the algorithm library, and calculates the fitness index of the algorithm on the feature space description matrix; Division module: Construct an optimization objective function based on the complexity index and the fitness index, select the optimal algorithm by solving the optimization objective function, use the selected algorithm to train the feature matrix, generate a risk control model, and divide users into different groups and generate attribute labels based on the output of the risk control model; Risk warning module: Based on the current attribute tags, a risk control rule set is constructed to conduct risk assessment on each new transaction and calculate the risk score. When the risk score is less than the preset threshold, a risk warning is issued and a risk warning report is generated.

[0029] It should be noted that the risk warning system based on adaptive algorithm selection provided in the above embodiment and the risk warning method based on adaptive algorithm selection provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs operations has been described in detail in the method embodiment and will not be repeated here. In practical applications, the risk warning system based on adaptive algorithm selection provided in the above embodiment can allocate the above functions to different functional modules as needed, that is, divide the internal structure of the system into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0030] An embodiment of the present application also provides 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 electronic device implements the risk warning method based on adaptive algorithm selection provided in the above-mentioned embodiments.

[0031] The embodiment of the present application also provides a computer system of an electronic device. It should be noted that the computer system of an electronic device is only an example and should not bring any limitation to the function and scope of use of the embodiment of the present application.

[0032] Specifically, the computer system includes a central processing unit (CPU), which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) or the program loaded from the storage part to the random access memory (RAM), such as executing the method described in the above embodiment. In the RAM, various programs and data required for system operation are also stored. The CPU, ROM and RAM are connected to each other through a bus. The input / output (I / O) interface is also connected to the bus.

[0033] The following components are connected to the I / O interface: an input section including a keyboard, a mouse, etc.; an output section including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker; a storage section including a hard disk, etc.; and a communication section including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section performs communication processing via a network such as the Internet. A drive is also connected to the I / O interface as needed. Removable media, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., are installed on the drive as needed so that a computer program read therefrom is installed into the storage section as needed.

[0034] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication part, and / or installed from a removable medium. When the computer program is executed by a central processing unit (CPU), various functions defined in the system of the present application are executed.

[0035] It should be noted that the computer-readable medium shown in the embodiment of the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier, which carries a computer-readable computer program. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. A computer program contained on a computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0036] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. The names of these units do not, in some cases, constitute limitations on the units themselves.

[0037] Another aspect of the present application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the risk warning method based on adaptive algorithm selection as described above is implemented. The computer-readable storage medium can be included in the electronic device described in the above embodiment, or it can exist independently without being assembled into the electronic device.

[0038] Another aspect of the present application also provides a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device reads the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the risk warning method based on adaptive algorithm selection provided in each of the above embodiments.

[0039] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. The risk early warning method based on adaptive algorithm selection is characterized by: include: S1: extracting historical integral data from the integral system database, constructing an original data set, and preprocessing the original data set; S2: extracting basic parameters from the preprocessed original data set, and constructing comprehensive risk characteristic parameters based on the basic parameters; S3: Define an algorithm library containing multiple machine learning algorithms, construct a feature space description matrix based on basic parameters and comprehensive risk feature parameters, calculate the complexity index of each algorithm in the algorithm library, and calculate the fitness index of the algorithm on the feature space description matrix; S4: Construct an optimization objective function based on the complexity index and the fitness index, select the optimal algorithm by solving the optimization objective function, use the selected algorithm to train the feature matrix, generate a risk control model, and divide users into different groups and generate attribute labels based on the output of the risk control model; S5: Based on the current attribute labels, a risk control rule set is constructed to conduct risk assessment on each new transaction and calculate the risk score. When the risk score is less than the preset threshold, a risk warning is issued and a risk warning report is generated.

2. The risk early warning method based on adaptive algorithm selection according to claim 1 is characterized in that: The basic parameters include user information and transaction information. The transaction information includes region code, transaction time interval, average transaction time interval, number of transaction location changes, total number of device types and frequency of user device use, points change amount, points change type and points acquisition and use ratio.

3. The risk early warning method based on adaptive algorithm selection according to claim 2 is characterized in that: The comprehensive risk characteristic parameters include spatiotemporal volatility characteristics, transaction pattern diversity characteristics and point utilization characteristics; The spatiotemporal volatility characteristics are calculated based on the trading time interval, the average trading time interval, and the number of trading location changes; Calculate transaction pattern diversity characteristics based on the total number of device types and user device usage frequency; The integral utilization characteristics are calculated based on the integral change amount, integral change type, and integral acquisition and utilization ratio.

4. The risk early warning method based on adaptive algorithm selection according to claim 3 is characterized in that: The calculation logic of the spatiotemporal volatility feature is: ,in, For users The spatiotemporal volatility characteristics of For users In the The time interval between the previous transaction and the next transaction. For users The average transaction time interval, For users The total number of transactions, is the number of transaction location changes, is the maximum number of transaction location changes, is the power exponent parameter, To adjust the parameters; The calculation logic of the transaction pattern diversity feature is: ,in, For users The diversity of transaction patterns, is the total number of device types, For users No. The frequency of user device usage, and is the power exponent parameter, is the total number of time periods in a day, For the The proportion of transactions conducted in a time period; The calculation logic of the points utilization feature is: ,in, For users The points utilization feature, For users In the The change in points for each transaction, is the type of integral change, coded as a categorical variable, The ratio of points earned and used is: is the power exponent parameter, To adjust the parameters.

5. The risk early warning method based on adaptive algorithm selection according to claim 1 is characterized in that: The feature space description matrix is ​​constructed based on the basic parameters and comprehensive risk characteristic parameters, including: Construct the initial feature vector by using the basic parameters and comprehensive risk characteristic parameters in any order ,in, , is the number of basic parameters and comprehensive risk characteristic parameters, For the basic parameters or comprehensive risk characteristic parameters; therefore, the initial feature space description matrix ,and , is the number of users, is a mathematical symbol representing the set of real numbers; The initial feature space description matrix Convert to polynomial feature space description matrix , , where the basis function Defined as: , is the total number of polynomial features, is the initial eigenvector Middle Features, is a polynomial basis function For The power of features, and satisfies ; The initial feature space description matrix and the polynomial feature space description matrix splicing, and generate the final feature space description matrix through principal component analysis and dimensionality reduction, where , is the final feature space description matrix, For principal component analysis, is the feature weight matrix after dimensionality reduction.

6. The risk early warning method based on adaptive algorithm selection according to claim 5 is characterized in that: The calculation logic of the complexity index of each algorithm in the algorithm library is: ,in, For the The complexity index of an algorithm is is the number of feature dimensions of the feature weight matrix after dimensionality reduction, For the The importance weight of the feature, For the The algorithm for The sensitivity of a feature, No. The discreteness of the features, To adjust the parameters; The calculation logic of the algorithm's fitness index on the feature space description matrix is: according to the final feature space description matrix Constructing the kernel matrix , ,in, is a similarity evaluation function used to evaluate the similarity between users. ,in, For the The fitness index of an algorithm is For users and users The similarity evaluation function value of For the The algorithm is used by the user and users The similarity evaluation value of .

7. The risk early warning method based on adaptive algorithm selection according to claim 6 is characterized in that: In step S4, the optimization objective function is: ,in, To optimize the objective function, and is the weight coefficient, and the sum is 1; Select the algorithm that optimizes the minimum value of the objective function , according to the algorithm Train the feature matrix to generate a risk control model. Based on the output of the risk control model, divide users into different groups and generate attribute labels. In step S5, the risk control rule set is a set of multiple parameter judgments, wherein each parameter is set with a preset value or interval; Get the attribute label of the user of each new transaction, get the risk control rule set of the attribute label, judge the parameters of the new transaction, when the conditions are met, the score value of the parameter is set to 1, when the conditions are not met, the score value of the parameter is set to 0, and the weighted sum of all parameters of the new transaction is obtained to obtain the risk score.

8. A risk warning system based on adaptive algorithm selection, used to implement the risk warning method based on adaptive algorithm selection according to any one of claims 1 to 7, characterized in that: include: Data extraction module: extracting historical integral data from the integral system database, constructing an original data set, and preprocessing the original data set; Feature construction module: extracting basic parameters from the preprocessed original data set and constructing comprehensive risk feature parameters based on the basic parameters; Calculation module: defines an algorithm library containing multiple machine learning algorithms, constructs a feature space description matrix based on basic parameters and comprehensive risk feature parameters, calculates the complexity index of each algorithm in the algorithm library, and calculates the fitness index of the algorithm on the feature space description matrix; Division module: Construct an optimization objective function based on the complexity index and the fitness index, select the optimal algorithm by solving the optimization objective function, use the selected algorithm to train the feature matrix, generate a risk control model, and divide users into different groups and generate attribute labels based on the output of the risk control model; Risk warning module: Based on the current attribute tags, a risk control rule set is constructed to conduct risk assessment on each new transaction and calculate the risk score. When the risk score is less than the preset threshold, a risk warning is issued and a risk warning report is generated.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the risk warning method based on adaptive algorithm selection as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is caused to execute the risk warning method based on adaptive algorithm selection as described in any one of claims 1 to 7.

Images