Risk identification method and device based on random forest algorithm, equipment and medium
Through the risk identification method based on the random forest algorithm, the decision tree is trained using historical user risk indicators, and the problem of unsatisfactory risk user identification efficiency in mobile payment services is solved, achieving more accurate and efficient risk identification.
Patent Information
- Application Number
- CN202411540808.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-30
- Publication Date
- 2025-03-04
AI Technical Summary
The prior art is not effective in identifying risk users in complex aggregation payment services in mobile payment services, and fails to form a risk monitoring mechanism based on the service itself.
A risk identification method based on a random forest algorithm is adopted to obtain historical user risk indicators for preprocessing, generate historical risk training sets, and use these data to train the target decision tree to automatically extract risk indicators from historical data and conduct risk prediction.
It improves the accuracy of identifying user risks, reduces risk assessment errors caused by human factors, and achieves more efficient risk user identification.
Smart Images
Figure CN119557773A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data analysis technology, which is applied to the field of financial technology business, and in particular to a risk identification method, device, equipment and medium based on a random forest algorithm. Background Art
[0002] With the rise of mobile, commercial banks have also expanded their mobile payment business. Mobile payment refers to electronic currency payment made by consumers using mobile phones and other electronic products. It is mainly centered on mobile terminals and uses mobile terminals to settle and pay for purchased products. With the help of mobile payment technology, consumers no longer need to use cash, but can purchase a series of goods, virtual goods or services by using only a mobile device such as a smartphone. Mobile payment can be regarded as the continuation and upgrade of the traditional POS (Point of Sale) acquiring business. Compared with cash and traditional card swiping machines or POS machine payment methods, barcode payment is faster and more friendly.
[0003] With the development of commercial banks' mobile payment business, various risks have begun to attract attention. With the intensive introduction of policies by relevant regulatory authorities and the continuous improvement of the enforcement of regulatory agencies, commercial banks have increased their investment in risk management and established their own risk control models. Since mobile payment business is essentially the same as traditional POS business, some commercial banks directly apply the risk control strategy of traditional POS merchants, but the efficiency of identifying risky merchants is not ideal, especially for more complex aggregate payment business; some commercial banks or payment institutions use the risk warnings of third-party payment giants or anti-money laundering departments as a criterion, but in fact they have not formed a risk monitoring mechanism based on the business itself in accordance with regulatory requirements. Therefore, how to improve the accuracy of identifying user risks through artificial intelligence technology has become a technical problem that needs to be solved urgently. Summary of the invention
[0004] The present application provides a risk identification method, apparatus, device and medium based on a random forest algorithm to improve the accuracy of identifying user risks through artificial intelligence technology.
[0005] In a first aspect, the present application provides a risk identification method based on a random forest algorithm, the method comprising:
[0006] Obtaining historical user risk indicators, and preprocessing the historical user risk indicators to generate a historical risk training set;
[0007] Generating a target decision tree according to the historical risk training set;
[0008] The target decision tree is used to perform risk prediction on all current user risk indicators, generate risk prediction results, and identify risky users based on the risk prediction results.
[0009] In a second aspect, the present application also provides a risk identification device based on a random forest algorithm, the device comprising:
[0010] A historical risk training set generation module is used to obtain historical user risk indicators and pre-process the historical user risk indicators to generate a historical risk training set;
[0011] A target decision tree generation module, used to generate a target decision tree according to the historical risk training set;
[0012] The risk identification module is used to perform risk prediction on all current user risk indicators through the target decision tree, generate risk prediction results and identify risky users based on the risk prediction results.
[0013] In a third aspect, the present application also provides a computer device, comprising a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and implement the risk identification method based on the random forest algorithm as described above when executing the computer program.
[0014] In a fourth aspect, the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the processor implements the risk identification method based on the random forest algorithm as described above.
[0015] The present application discloses a risk identification method, device, equipment and medium based on a random forest algorithm. The risk identification method based on a random forest algorithm includes obtaining historical user risk indicators, preprocessing the historical user risk indicators, generating a historical risk training set; generating a target decision tree based on the historical risk training set; performing risk prediction on all current user risk indicators through the target decision tree, generating risk prediction results and identifying risky users based on the risk prediction results. In the above manner, the present application generates a training set using historical user risk indicators, and uses these data to train a target decision tree, automatically extracts risk indicators from historical data, generates a training set, and uses a machine learning model to perform risk prediction, thereby reducing risk assessment errors caused by human factors, thereby improving the accuracy of identifying user risks. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0017] Figure 1 is a schematic flow chart of a risk identification method based on a random forest algorithm provided in the first embodiment of the present application;
[0018] Figure 2 is a schematic flow chart of a risk identification method based on a random forest algorithm provided in the second embodiment of the present application;
[0019] Figure 3 A schematic block diagram of a risk identification device based on a random forest algorithm provided in an embodiment of the present application;
[0020] Figure 4 A schematic block diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0021] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0022] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.
[0023] It should be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0024] It should also be understood that the term “and / or” used in the specification and appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0025] The embodiments of the present application provide a risk identification method, device, equipment and medium based on a random forest algorithm. Among them, the risk identification method based on a random forest algorithm can be applied to a server, by using historical user risk indicators to generate a training set, and using these data to train a target decision tree, automatically extracting risk indicators from historical data, generating a training set, and using a machine learning model to predict risks, thereby reducing risk assessment errors caused by human factors, thereby improving the accuracy of identifying user risks. Among them, the server can be an independent server or a server cluster.
[0026] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.
[0027] See also Figure 1 , Figure 1 This is a schematic flow chart of a risk identification method based on a random forest algorithm provided by the first embodiment of the present application. The risk identification method based on a random forest algorithm can be applied to a server to generate a training set by using historical user risk indicators, and use these data to train a target decision tree, automatically extract risk indicators from historical data, generate a training set, and use a machine learning model to predict risks, thereby reducing risk assessment errors caused by human factors and improving the accuracy of identifying user risks.
[0028] like Figure 1 As shown, the risk identification method based on the random forest algorithm specifically includes steps S10 to S30.
[0029] Step S10: Obtain historical user risk indicators, and pre-process the historical user risk indicators to generate a historical risk training set;
[0030] Step S20: generating a target decision tree according to the historical risk training set;
[0031] Step S30: perform risk prediction on all current user risk indicators through the target decision tree, generate risk prediction results, and identify risky users based on the risk prediction results.
[0032] In one embodiment, the training method of generating a target decision tree through a historical risk training set in this embodiment can be implemented in the following manner:
[0033] 1. In the historical risk training set, the Bootstrap method is used to randomly select k new sample sets (historical risk data sets) with replacement. The k sample sets (historical risk data sets) correspond to k classification (regression) trees respectively; the samples that are still not selected after k times constitute the out-of-bag data. The basic steps of the Bootstrap method are as follows:
[0034] Resampling: Randomly extract samples of the same size as the original sample from the original data set with replacement. This process is repeated many times (for example, 1,000 times or more) to form multiple "bootstrap samples". Calculate statistics: Calculate the statistics of interest (for example, mean, median, variance, regression coefficient, etc.) for each bootstrap sample.
[0035] Construct a distribution: Construct an empirical distribution of the statistic based on all the statistics calculated from the bootstrap samples. Estimate parameters: From this distribution, you can estimate the statistical properties of the original statistic, such as the standard error, confidence interval, bias, kurtosis, etc.
[0036] 2. Suppose there are N features; at each node of each decision tree, M (M≤N) features are randomly selected; according to the selected node splitting criterion, one of the most representative features is selected for node splitting.
[0037] 3. Each tree grows to its maximum without any pruning.
[0038] 4. All decision trees constitute a random forest, and the results are calculated by simple voting or averaging as the classification results of the entire forest.
[0039] Specifically, k samples are randomly selected with replacement from the original training data set to form the first Bootstrap sample set. This process is repeated k times, and each extraction forms a new Bootstrap sample set, and each sample set is used to build a decision tree.
[0040] For each decision tree, at each node, a subset of features is randomly selected, and then the best split feature is selected for splitting, allowing each tree to grow fully without pruning.
[0041] For classification problems, each tree makes a class prediction. The final prediction of a random forest is the majority vote of all the tree predictions. For regression problems, each tree makes a numerical prediction. The final prediction of a random forest is the average of all the tree predictions.
[0042] This embodiment discloses a risk identification method, device, equipment and medium based on a random forest algorithm. The risk identification method based on a random forest algorithm includes obtaining historical user risk indicators, preprocessing the historical user risk indicators, generating a historical risk training set; generating a target decision tree based on the historical risk training set; performing risk prediction on all current user risk indicators through the target decision tree, generating risk prediction results and identifying risky users based on the risk prediction results. In the above manner, the present application generates a training set using historical user risk indicators, and uses these data to train a target decision tree, automatically extracts risk indicators from historical data, generates a training set, and uses a machine learning model to perform risk prediction, thereby reducing risk assessment errors caused by human factors, thereby improving the accuracy of identifying user risks.
[0043] See also Figure 2 , Figure 2 This is a schematic flow chart of a risk identification method based on a random forest algorithm provided in the second embodiment of the present application. The risk identification method based on a random forest algorithm can be applied to a server to generate a training set by using historical user risk indicators, and use these data to train a target decision tree, automatically extract risk indicators from historical data, generate a training set, and use a machine learning model to predict risks, thereby reducing risk assessment errors caused by human factors and improving the accuracy of identifying user risks.
[0044] based on Figure 1 The embodiment shown, this embodiment Figure 2 As shown, step S20 includes steps S201 to S203.
[0045] Step S201: determining a sample subspace from the historical risk training set, wherein the sample subspace includes at least one type of characteristic sample information;
[0046] Step S202: extract at least one set of the characteristic sample information from the sample subspace, and iteratively process each set of the characteristic sample information;
[0047] Step S203: when the iteration of each feature sample information stops, generating the target decision tree based on the iteratively processed feature sample information and each sample subspace.
[0048] In one embodiment, a sample subspace is randomly selected from the historical risk training set. This subspace includes samples selected from the original data set, which will be used to build a decision tree, ensuring that the sample subspace contains at least one feature sample information, that is, at least one feature used to train the model.
[0049] Extract a set of feature sample information from the sample subspace. This set of information will be used to build a decision tree. If necessary, repeat this step to extract multiple sets of feature sample information from the sample subspace, each of which is used to build a different decision tree. Perform iterative processing on each set of extracted feature sample information. Iterative processing may include but is not limited to feature selection, feature transformation, missing value processing, etc. During the iterative processing, a splitting criterion (such as information gain, Gini impurity) is applied to decide how to split the node. Continue iterative processing until the stopping condition is met. The stopping condition may include reaching the maximum depth of the tree, the number of samples in the node is lower than a certain threshold, or the purity of the node is high enough.
[0050] Once the iterative processing of all feature sample information stops, use the processed data to generate the target decision tree, ensuring that each tree grows to the maximum without any pruning to retain as much information as possible. For classification problems, each leaf node represents a category; for regression problems, each leaf node represents a predicted value. If it is a single decision tree in a random forest, this step may not apply. But for the entire random forest, the results of all individual decision trees need to be aggregated.
[0051] The present embodiment discloses a risk identification method, device, equipment and medium based on a random forest algorithm, wherein the risk identification method based on a random forest algorithm includes obtaining historical user risk indicators, preprocessing the historical user risk indicators, and generating a historical risk training set; determining a sample subspace from the historical risk training set, wherein the sample subspace includes at least one feature sample information; extracting at least one set of the feature sample information from the sample subspace, and iteratively processing each of the feature sample information; when each of the feature sample information stops iterating, generating the target decision tree based on each of the iteratively processed feature sample information and each of the sample subspaces; performing risk prediction on all current user risk indicators through the target decision tree, generating risk prediction results and identifying risk users according to the risk prediction results. In the above manner, the present application generates a training set by using historical user risk indicators, and uses these data to train a target decision tree, automatically extracts risk indicators from historical data, generates a training set, and uses a machine learning model for risk prediction, thereby reducing risk assessment errors caused by human factors, thereby improving the accuracy of identifying user risks.
[0052] based on Figure 2 In the embodiment shown, in this embodiment, step S201 includes:
[0053] Randomly extracting a first preset number of feature sample information from the historical risk training set by a preset bootstrap method;
[0054] Taking a first preset number of feature sample information as a first feature node;
[0055] Each of the first characteristic nodes and the node characteristic information contained in each of the first characteristic nodes are determined as the sample subspace.
[0056] In one embodiment, a first preset number is determined, that is, the number of feature sample information that is expected to be extracted from the historical risk training set. This number is usually related to the size of the training set and may be a ratio of the size of the training set or a fixed number.
[0057] Use the bootstrap method to randomly extract the first preset number of feature sample information from the historical risk training set. This process is with replacement, which means that the same sample may be extracted multiple times, and some samples may not be extracted once, ensuring that each extraction is independent to ensure the randomness of the sample.
[0058] Each extracted feature sample information is taken as a node, which is called the first feature node. Each first feature node represents a sample in the training set and contains all the feature information of the sample. All the first feature nodes and the feature information contained in these nodes are combined to determine the sample subspace. This sample subspace will be used for subsequent model training, such as building a decision tree or other machine learning model.
[0059] The present embodiment discloses a risk identification method, apparatus, device and medium based on a random forest algorithm. The risk identification method based on a random forest algorithm includes obtaining historical user risk indicators, and preprocessing the historical user risk indicators to generate a historical risk training set; randomly extracting a first preset number of feature sample information from the historical risk training set through a preset self-help method; using each of the first preset number of feature sample information as a first feature node; determining each of the first feature nodes and the node feature information contained in each of the first feature nodes as the sample subspace; extracting at least one group of the feature sample information from the sample subspace, and iteratively processing each of the feature sample information; when each of the feature sample information stops iterating, generating the target decision tree based on each of the iteratively processed feature sample information and each of the sample subspaces; performing risk prediction on all current user risk indicators through the target decision tree, generating a risk prediction result and identifying risky users based on the risk prediction result. Through the above method, the present application generates a training set by using historical user risk indicators, and uses these data to train the target decision tree, automatically extracts risk indicators from historical data, generates a training set, and uses a machine learning model for risk prediction, thereby reducing risk assessment errors caused by human factors and improving the accuracy of identifying user risks.
[0060] In a more preferred embodiment, based on Figure 2 In the embodiment shown, in this embodiment, step S202 includes:
[0061] Extracting a second preset number of the feature sample information from each of the feature sample information in the sample subspace by using the preset self-help method, wherein the second preset number is smaller than the first preset number;
[0062] Determining the second preset number of feature sample information as second feature nodes;
[0063] Node feature information contained in the second feature node is obtained, and each node feature information in the second feature node is iteratively processed.
[0064] In one embodiment, a second preset number is determined, i.e., the number of feature sample information that is expected to be extracted from the sample subspace, which should be less than the first preset number. The second preset number of feature sample information is randomly extracted from the sample subspace using a bootstrap method. This process is with replacement, which means that the same sample may be extracted multiple times, and some samples may not be extracted once, ensuring that each extraction is independent to ensure the randomness of the sample.
[0065] Each extracted feature sample information is taken as a node, which are called second feature nodes. Each second feature node represents a sample in the sample subspace and contains all feature information of the sample. The total number of second feature nodes is equal to the second preset number.
[0066] The iterative process continues until a stopping condition is met, which may include reaching the maximum depth of the tree, the number of samples in a node falling below a certain threshold, or the purity of the node being high enough.
[0067] The present embodiment discloses a risk identification method, apparatus, device and medium based on a random forest algorithm. The risk identification method based on a random forest algorithm includes obtaining historical user risk indicators, preprocessing the historical user risk indicators, and generating a historical risk training set; determining a sample subspace from the historical risk training set, wherein the sample subspace includes at least one feature sample information; extracting a second preset number of feature sample information from each feature sample information in the sample subspace by the preset self-help method, wherein the second preset number is less than the first preset number; determining the second preset number of feature sample information as a second feature node; obtaining node feature information contained in the second feature node, and iteratively processing each node feature information in the second feature node; when each feature sample information stops iterating, generating the target decision tree based on each iteratively processed feature sample information and each sample subspace; performing risk prediction on all current user risk indicators by the target decision tree, generating a risk prediction result, and identifying risky users based on the risk prediction result. Through the above method, the present application generates a training set by using historical user risk indicators, and uses these data to train the target decision tree, automatically extracts risk indicators from historical data, generates a training set, and uses a machine learning model for risk prediction, thereby reducing risk assessment errors caused by human factors and improving the accuracy of identifying user risks.
[0068] In a more preferred embodiment, before step S30, the following steps are included:
[0069] Detecting whether other feature nodes corresponding to each of the feature sample information stop iterating;
[0070] When it is detected that the node characteristic information of the other characteristic nodes stops growing, it is determined that the characteristic sample information stops iterating.
[0071] based on Figure 1 In the embodiment shown, in this embodiment, step S30 includes:
[0072] Determining a preset risk threshold based on each of the current user risk indicators;
[0073] Calculate the individual risk values corresponding to the current user risk indicators through the target decision tree and the current user risk indicators, wherein the risk prediction result includes the individual risk values;
[0074] Based on the preset weights and the individual risk values, a comprehensive risk value corresponding to all the current user risk indicators is calculated, wherein the risk prediction result includes the comprehensive risk value;
[0075] The user whose comprehensive risk value is greater than or equal to the preset risk threshold is determined as the risky user.
[0076] Specifically, based on historical data, business requirements, and risk management strategies, one or more preset risk thresholds are determined to classify users as risky users or non-risky users.
[0077] Use the target decision tree to evaluate the risk indicators of each current user. For each user, the decision tree will calculate a single risk value based on its risk indicator. This value represents the risk level of the user under a single risk indicator. Summarize the single risk values of each user to form the risk prediction result for that user. This result includes all single risk values, and assigns a preset weight to each single risk value. These weights reflect the importance of different risk indicators in the overall risk assessment. Multiply each single risk value by its corresponding weight, and then add these weighted values to get the comprehensive risk value of each user.
[0078] The comprehensive risk value of each user is compared with a preset risk threshold. If the comprehensive risk value of the user is greater than or equal to the preset risk threshold, the user is determined to be a risky user.
[0079] Based on any of the above embodiments, in this embodiment, step S10 includes:
[0080] Obtaining certificate information, enterprise information, geographic information, industry information and / or transaction information as the historical user risk indicator;
[0081] Acquire historical risk data, and classify and process each of the historical user risk indicators based on the historical risk data;
[0082] The type of each of the historical user risk indicators is used as a sample label, and each of the sample labels and each of the historical user risk indicators is used as the historical risk training set.
[0083] See also Figure 3 , Figure 3 The embodiment of the present application provides a schematic block diagram of a risk identification device based on a random forest algorithm, and the risk identification device based on a random forest algorithm is used to execute the aforementioned risk identification method based on a random forest algorithm. The risk identification device based on a random forest algorithm can be configured on a server.
[0084] like Figure 3 As shown, the risk identification device 400 based on the random forest algorithm includes:
[0085] A historical risk training set generation module 410 is used to obtain historical user risk indicators and pre-process the historical user risk indicators to generate a historical risk training set;
[0086] A target decision tree generation module 420 is used to generate a target decision tree according to the historical risk training set;
[0087] The risk identification module 430 is used to perform risk prediction on all current user risk indicators through the target decision tree, generate risk prediction results and identify risky users based on the risk prediction results.
[0088] Furthermore, the target decision tree generation module 420 includes:
[0089] A sample subspace determining unit, configured to determine a sample subspace from the historical risk training set, wherein the sample subspace includes at least one type of characteristic sample information;
[0090] An iterative unit, used for extracting at least one set of the characteristic sample information from the sample subspace, and iteratively processing each set of the characteristic sample information;
[0091] The target decision tree generating unit is used to generate the target decision tree based on the iteratively processed feature sample information and the sample subspaces when the iteration of the feature sample information is stopped.
[0092] Furthermore, the sample subspace determination unit includes:
[0093] A feature sample information extraction subunit, used for randomly extracting a first preset amount of the feature sample information from the historical risk training set by a preset bootstrap method;
[0094] A first feature node determination subunit, configured to use a first preset number of feature sample information as a first feature node;
[0095] The sample subspace determination subunit is used to determine each of the first characteristic nodes and the node characteristic information contained in each of the first characteristic nodes as the sample subspace.
[0096] Furthermore, the iteration unit includes:
[0097] a feature sample information extraction subunit, configured to extract a second preset number of the feature sample information from each of the feature sample information in the sample subspace by using the preset self-help method, wherein the second preset number is smaller than the first preset number;
[0098] A second characteristic node determination subunit, configured to determine the characteristic sample information of the second preset number as a second characteristic node;
[0099] The iterative processing subunit is used to obtain the node feature information contained in the second feature node and iteratively process each node feature information in the second feature node.
[0100] Furthermore, the risk identification device 400 based on the random forest algorithm includes:
[0101] An iteration detection module, used to detect whether other feature nodes corresponding to each of the feature sample information stop iterating;
[0102] The iteration stop detection module is used to determine that the feature sample information stops iterating when it is detected that the node feature information of the other feature nodes stops growing.
[0103] Furthermore, the risk identification module 430 includes:
[0104] A preset risk threshold determination unit, used to determine a preset risk threshold based on each of the current user risk indicators;
[0105] A single risk value determination unit, configured to calculate the single risk value corresponding to each current user risk index through the target decision tree and each current user risk index, wherein the risk prediction result includes the single risk value;
[0106] a comprehensive risk value determination unit, configured to calculate a comprehensive risk value corresponding to all the current user risk indicators based on a preset weight and each of the individual risk values, wherein the risk prediction result includes the comprehensive risk value;
[0107] The risk identification unit is used to determine the users whose comprehensive risk value is greater than or equal to the preset risk threshold as the risky users.
[0108] It should be noted that those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device and each module can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.
[0109] The above-mentioned device can be implemented in the form of a computer program. Figure 4 Runs on the computer device shown.
[0110] See also Figure 4 , Figure 4 1 is a schematic block diagram of a computer device provided in an embodiment of the present application. The computer device may be a server.
[0111] See also Figure 4The computer device includes a processor, a memory and a network interface connected through a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.
[0112] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any risk identification method based on the random forest algorithm.
[0113] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.
[0114] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any risk identification method based on the random forest algorithm.
[0115] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0116] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.
[0117] In one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:
[0118] Obtaining historical user risk indicators, and preprocessing the historical user risk indicators to generate a historical risk training set;
[0119] Generating a target decision tree according to the historical risk training set;
[0120] The target decision tree is used to perform risk prediction on all current user risk indicators, generate risk prediction results, and identify risky users based on the risk prediction results.
[0121] In one embodiment, a target decision tree is generated based on the historical risk training set to achieve:
[0122] Determining a sample subspace from the historical risk training set, wherein the sample subspace includes at least one type of characteristic sample information;
[0123] Extracting at least one group of the characteristic sample information from the sample subspace, and iteratively processing each of the characteristic sample information;
[0124] In the case where iteration of each of the feature sample information stops, the target decision tree is generated based on the iteratively processed feature sample information and each of the sample subspaces.
[0125] In one embodiment, a sample subspace is determined from the historical risk training set, wherein the sample subspace includes at least one type of characteristic sample information, for achieving:
[0126] Randomly extracting a first preset number of feature sample information from the historical risk training set by a preset bootstrap method;
[0127] Taking a first preset number of feature sample information as a first feature node;
[0128] Each of the first characteristic nodes and the node characteristic information contained in each of the first characteristic nodes are determined as the sample subspace.
[0129] In one embodiment, at least one set of the characteristic sample information is extracted from the sample subspace, and each set of the characteristic sample information is iteratively processed to achieve:
[0130] Extracting a second preset number of the feature sample information from each of the feature sample information in the sample subspace by using the preset self-help method, wherein the second preset number is smaller than the first preset number;
[0131] Determining the second preset number of feature sample information as second feature nodes;
[0132] Node feature information contained in the second feature node is obtained, and each node feature information in the second feature node is iteratively processed.
[0133] In one embodiment, when the iteration of each feature sample information stops, before the target decision tree is generated based on the iteratively processed feature sample information and each sample subspace, it is used to implement:
[0134] Detecting whether other feature nodes corresponding to each of the feature sample information stop iterating;
[0135] When it is detected that the node characteristic information of the other characteristic nodes stops growing, it is determined that the characteristic sample information stops iterating.
[0136] In one embodiment, risk prediction is performed on all current user risk indicators through the target decision tree, risk prediction results are generated, and risky users are identified based on the risk prediction results, so as to achieve:
[0137] Determining a preset risk threshold based on each of the current user risk indicators;
[0138] Calculate the individual risk values corresponding to the current user risk indicators through the target decision tree and the current user risk indicators, wherein the risk prediction result includes the individual risk values;
[0139] Based on the preset weights and the individual risk values, a comprehensive risk value corresponding to all the current user risk indicators is calculated, wherein the risk prediction result includes the comprehensive risk value;
[0140] The user whose comprehensive risk value is greater than or equal to the preset risk threshold is determined as the risky user.
[0141] In one embodiment, historical user risk indicators are obtained and preprocessed to generate a historical risk training set for achieving:
[0142] Obtaining certificate information, enterprise information, geographic information, industry information and / or transaction information as the historical user risk indicator;
[0143] Acquire historical risk data, and classify and process each of the historical user risk indicators based on the historical risk data;
[0144] The type of each of the historical user risk indicators is used as a sample label, and each of the sample labels and each of the historical user risk indicators is used as the historical risk training set.
[0145] A computer-readable storage medium is also provided in an embodiment of the present application, wherein the computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and the processor executes the program instructions to implement any risk identification method based on a random forest algorithm provided in an embodiment of the present application.
[0146] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiment, such as a hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart memory card (Smart Medi a Card, SMC), a secure digital (Secure Digital, SD) card, a flash memory card (Flash Card), etc., equipped on the computer device.
[0147] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present application, and these modifications or replacements should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.
Claims
1. A risk identification method based on random forest algorithm, characterized in that: include: Obtaining historical user risk indicators, and preprocessing the historical user risk indicators to generate a historical risk training set; Generating a target decision tree according to the historical risk training set; The target decision tree is used to perform risk prediction on all current user risk indicators, generate risk prediction results, and identify risky users based on the risk prediction results.
2. The risk identification method based on random forest algorithm according to claim 1 is characterized in that: Generating a target decision tree according to the historical risk training set includes: Determining a sample subspace from the historical risk training set, wherein the sample subspace includes at least one type of characteristic sample information; Extracting at least one group of the characteristic sample information from the sample subspace, and iteratively processing each of the characteristic sample information; In the case where iteration of each of the feature sample information stops, the target decision tree is generated based on the iteratively processed feature sample information and each of the sample subspaces.
3. The risk identification method based on random forest algorithm according to claim 2 is characterized in that: The determining of a sample subspace from the historical risk training set, wherein the sample subspace includes at least one type of characteristic sample information, including: Randomly extracting a first preset number of feature sample information from the historical risk training set by a preset bootstrap method; Taking a first preset number of feature sample information as a first feature node; Each of the first characteristic nodes and the node characteristic information contained in each of the first characteristic nodes are determined as the sample subspace.
4. The risk identification method based on random forest algorithm according to claim 3 is characterized in that: The extracting at least one group of the feature sample information from the sample subspace, and iteratively processing each of the feature sample information, comprises: Extracting a second preset number of the feature sample information from each of the feature sample information in the sample subspace by using the preset self-help method, wherein the second preset number is smaller than the first preset number; Determining the second preset number of feature sample information as second feature nodes; Node feature information contained in the second feature node is obtained, and each node feature information in the second feature node is iteratively processed.
5. The risk identification method based on random forest algorithm according to claim 4 is characterized in that: In the case where the iteration of each feature sample information stops, before generating the target decision tree based on each feature sample information and each sample subspace processed by iteration, the method includes: Detecting whether other feature nodes corresponding to each of the feature sample information stop iterating; When it is detected that the node characteristic information of the other characteristic nodes stops growing, it is determined that the characteristic sample information stops iterating.
6. The risk identification method based on random forest algorithm according to claim 1 is characterized in that: The step of performing risk prediction on all current user risk indicators through the target decision tree, generating risk prediction results, and identifying risky users according to the risk prediction results includes: Determining a preset risk threshold based on each of the current user risk indicators; Calculate the individual risk values corresponding to the individual risk indicators of the current users through the target decision tree and the individual risk indicators of the current users, wherein the risk prediction result includes the individual risk values; Based on the preset weights and the individual risk values, a comprehensive risk value corresponding to all the current user risk indicators is calculated, wherein the risk prediction result includes the comprehensive risk value; The user whose comprehensive risk value is greater than or equal to the preset risk threshold is determined as the risky user.
7. The risk identification method based on random forest algorithm according to any one of claims 1 to 6, characterized in that: The obtaining of historical user risk indicators and preprocessing of the historical user risk indicators to generate a historical risk training set includes: Obtaining certificate information, enterprise information, geographic information, industry information and / or transaction information as the historical user risk indicator; Acquire historical risk data, and classify and process each of the historical user risk indicators based on the historical risk data; The type of each of the historical user risk indicators is used as a sample label, and each of the sample labels and each of the historical user risk indicators is used as the historical risk training set.
8. A risk identification device based on random forest algorithm, characterized in that: include: A historical risk training set generation module is used to obtain historical user risk indicators and pre-process the historical user risk indicators to generate a historical risk training set; A target decision tree generation module, used to generate a target decision tree according to the historical risk training set; The risk identification module is used to perform risk prediction on all current user risk indicators through the target decision tree, generate risk prediction results and identify risky users based on the risk prediction results.
9. A computer device, characterized in that: The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is used to execute the computer program and implement the risk identification method based on the random forest algorithm as described in any one of claims 1 to 7 when executing the computer program.
10. 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, the processor implements the risk identification method based on the random forest algorithm according to any one of claims 1 to 7.