Risk control rule processing method and device based on random forest, equipment and medium

Abstract

The invention provides a risk control rule processing method and device based on a random forest, equipment and a medium. The method comprises the steps of obtaining historical credit data; performing chi-party box separation processing on the historical credit data to obtain multiple groups of chi-party box separation data; determining target feature data from the plurality of groups of chi-party binning data, wherein the target feature data is used for describing credit data corresponding to target credit features in the historical credit data; determining a target risk control rule from the risk control rule set based on the target feature data by adopting a random forest model; and performing data analysis on the target risk control rule to obtain a credit risk control rule, the credit risk control rule being used for detecting the credit risk of the loan person. Therefore, the collected historical credit data is subjected to automatic binning processing, and the random forest model is adopted to perform rule making on the target feature data obtained after binning, so that the credit risk control rule for detecting the credit risk of the loan personnel can be conveniently obtained, and the risk control efficiency is effectively improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the technical field of risk control rule processing, and in particular, to a method, apparatus, device and medium for risk control rule processing based on random forest. Background Art

[0002] With the rapid development of information technology, the demand for asset security in the digital age is increasing. When handling some loan businesses, it is necessary to conduct risk assessment on loan personnel in order to control their risks.

[0003] In the related art, risk control is usually performed by formulating risk control strategies. For example, professionals are required to manually bin loan data and prepare corresponding risk control strategies so that the prepared risk control strategies can perform risk control detection on loan personnel.

[0004] However, the risk control efficiency is low in the existing methods. Summary of the invention

[0005] The embodiments described herein provide a risk control rule processing method, apparatus, device and medium based on random forest, which overcome the above-mentioned problems.

[0006] In a first aspect, according to the content of the present disclosure, a risk control rule processing method based on random forest is provided, comprising:

[0007] Obtaining historical credit data, the historical credit data includes: pre-loan sample data, post-loan sample data and third-party sample data, the pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, the post-loan sample data is used to describe the credit data of the loan personnel after the loan, and the third-party sample data is used to describe the credit data stored in the third-party risk control platform;

[0008] Performing card-square binning processing on the historical credit data to obtain multiple groups of card-square binning data, each group of the card-square binning data including at least one of the following: the pre-loan sample data, the post-loan sample data, and the third-party sample data;

[0009] Determining target feature data from the plurality of groups of the chi-square binned data, the target feature data being used to describe the credit data corresponding to the target credit feature in the historical credit data;

[0010] Using a random forest model, a target risk control rule is determined from a risk control rule set based on the target feature data;

[0011] Data analysis is performed on the target risk control rules to obtain credit risk control rules, and the credit risk control rules are used to detect the credit risk of loan personnel.

[0012] In a second aspect, according to the content of the present disclosure, a risk control rule processing device based on random forest is provided, comprising:

[0013] An acquisition module is used to acquire historical credit data, wherein the historical credit data includes: pre-loan sample data, post-loan sample data and third-party sample data. The pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, the post-loan sample data is used to describe the credit data of the loan personnel after the loan, and the third-party sample data is used to describe the credit data stored in the third-party risk control platform;

[0014] A processing module, configured to perform card-square binning processing on the historical credit data to obtain multiple groups of card-square binning data, each group of the card-square binning data including at least one of the following: the pre-loan sample data, the post-loan sample data, and the third-party sample data;

[0015] A first determination module is used to determine target feature data from multiple groups of the chi-square binned data, wherein the target feature data is used to describe the credit data corresponding to the target credit feature in the historical credit data;

[0016] A second determination module is used to determine a target risk control rule from a risk control rule set based on the target feature data by using a random forest model;

[0017] The analysis module is used to perform data analysis on the target risk control rules to obtain credit risk control rules, and the credit risk control rules are used to detect the credit risk of loan personnel.

[0018] In a third aspect, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of the risk control rule processing method based on random forest in any of the above embodiments are implemented.

[0019] In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the risk control rule processing method based on random forest in any of the above embodiments are implemented.

[0020] The risk control rule processing method based on random forest provided in the embodiment of the present application obtains historical credit data, and the historical credit data includes: pre-loan sample data, post-loan sample data and third-party sample data, the pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, the post-loan sample data is used to describe the credit data of the loan personnel after the loan, and the third-party sample data is used to describe the credit data stored in the third-party risk control platform; the historical credit data is subjected to chi-square binning processing to obtain multiple groups of card-square binned data, and each group of card-square binned data includes at least one of the following: pre-loan sample data, post-loan sample data and third-party sample data; target feature data is determined from the multiple groups of card-square binned data, and the target feature data is used to describe the credit data corresponding to the target credit feature in the historical credit data; a random forest model is used to determine the target risk control rule from the risk control rule set based on the target feature data; data analysis is performed on the target risk control rule to obtain the credit risk control rule, and the credit risk control rule is used to detect the credit risk of the loan personnel. In this way, by automatically binning the summarized historical credit data and using the random forest model to formulate rules for the target feature data obtained after binning, it is easy to obtain credit risk control rules for detecting the credit risk of loan personnel, thereby effectively improving risk control efficiency.

[0021] The above description is only an overview of the technical solution of the embodiment of the present application. In order to more clearly understand the technical means of the embodiment 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 embodiment 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

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings of the embodiments will be briefly described below. It should be noted that the drawings described below only relate to some embodiments of the present disclosure, but are not intended to limit the present disclosure, wherein:

[0023] Figure 1 It is a flow chart of a risk control rule processing method based on random forest provided by the present invention.

[0024] Figure 2 It is a structural schematic diagram of a risk control rule processing device based on random forest provided by the present invention.

[0025] Figure 3 It is a structural schematic diagram of a computer device provided by the present disclosure.

[0026] It should be noted that the elements in the drawings are schematic and not drawn to scale. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the embodiments of the present disclosure clearer, the technical solution of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the described embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without creative work also fall within the scope of protection of the present disclosure.

[0028] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by a person skilled in the art to which the subject matter of the present disclosure belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the specification and the relevant art, and will not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, a statement that two or more parts are "connected" or "coupled" together shall mean that the parts are joined together directly or through one or more intermediate components.

[0029] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase "embodiments" in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0030] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists, A and B exist at the same time, and B exists. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. Terms such as "first" and "second" are only used to distinguish one component (or a part of a component) from another component (or another part of a component).

[0031] In the description of the present application, unless otherwise specified, "plurality" means more than two (including two), and similarly, "multiple groups" means more than two groups (including two).

[0032] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0033] Figure 1 is a flow chart of a risk control rule processing method based on random forest provided by an embodiment of the present disclosure, such as Figure 1As shown in the figure, the specific process of the risk control rule processing method based on random forest includes:

[0034] S110. Obtain historical credit data.

[0035] Among them, historical credit data includes: pre-loan sample data, post-loan sample data and third-party sample data. Pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, post-loan sample data is used to describe the credit data of the loan personnel after the loan, and third-party sample data is used to describe the credit data stored by the third-party risk control platform.

[0036] For example, pre-loan sample data may include but is not limited to: age value, vehicle age value, marital status, decision results (such as rejection of transfer to labor, etc.). Post-loan sample data may include but is not limited to: total loan periods, current loan period, overdue amount, repaid amount, due date and actual date of repayment, etc. Third-party sample data refers to some loan information recorded by loan personnel corresponding to pre-loan sample data and post-loan sample data on the third-party risk control platform.

[0037] After obtaining the historical credit data, the indicators of the historical credit data can be derived according to the time dimension to expand the diversity of the historical credit data. For example, the time dimension is the number of queries in the past x months, and combined with the operation dimension, such as the number of query institutions in the past x months, more indicator data can be obtained based on the number of queries in the past x months + the number of query institutions in the past x months to expand the historical credit data.

[0038] S120, performing chi-square binning processing on the historical credit data to obtain multiple groups of chi-square binning data.

[0039] Each group of card-square binned data includes at least one of the following: pre-loan sample data, post-loan sample data, and third-party sample data. That is, a group of card-square binned data includes: pre-loan sample data, or post-loan sample data, or post-loan sample data, or pre-loan sample data and post-loan sample data, or pre-loan sample data and third-party sample data, or post-loan sample data and third-party sample data, or pre-loan sample data, post-loan sample data, and third-party sample data.

[0040] In some embodiments, historical credit data is subjected to card-square binning processing to obtain multiple groups of card-square binning data, including:

[0041] Obtaining preset risk control indicators; based on the preset risk control indicators, obtaining risk control credit data from historical credit data, the risk control credit data including: multiple sub-credit data; based on the risk control similarity representation values ​​between the multiple sub-credit data, performing chi-square binning processing on the multiple sub-credit data to obtain multiple groups of chi-square binned data.

[0042] Among them, the preset risk control indicators are used to describe the overdue judgment indicators of credit data. That is, through the preset risk control indicators, bad sample data can be found from the historical credit data. The bad sample data can be determined based on the label rolling rate and vintage. In the historical credit data, each credit data will have a corresponding label (i.e. rolling rate and vintage). The rolling rate is M0 rolling to M1, M1 rolling to M2 and so on, to observe which interval is stable; vintage is to pull the post-loan data of different loan times to the same point in time to observe how many periods are stable; vintage can be used to regularly express how long the period is, and the rolling rate can be used to define bad samples. For example, overdue for more than 30 days in the past 6 months is overdue.

[0043] In some embodiments, before performing card-square binning processing on the historical credit data to obtain multiple groups of card-square binned data, the method further includes:

[0044] Acquire the credit data that meets the preset data type in the historical credit data, where the preset data type includes: data missing type and data abnormality type; perform data preprocessing on the credit data that meets the preset data type in the historical credit data to update the historical credit data;

[0045] Among them, data preprocessing includes: missing data filling and abnormal data replacement.

[0046] For example, if the first credit data in the historical credit data is data-missing type credit data, the first credit data is filled by the median / mean data of multiple credit data adjacent to the first credit data. If the second credit data in the historical credit data is data-abnormal type credit data, the second credit data is replaced by the median of multiple credit data adjacent to the second credit data.

[0047] Therefore, by filling in missing data and replacing abnormal data in the historical credit data that meets the preset data type, it is convenient to effectively ensure the data integrity of the historical credit data.

[0048] S130, determining target feature data from multiple groups of chi-square binned data.

[0049] The target feature data is used to describe the credit data corresponding to the target credit feature in the historical credit data. The target credit feature may be, for example, gender, position, etc.

[0050] In some embodiments, determining target feature data from multiple groups of chi-square binned data includes:

[0051] Obtain the bin risk control characterization value corresponding to each group of chi-square bin data; determine the chi-square bin data whose bin risk control characterization value is greater than the preset risk control characterization value as candidate feature data; determine the target feature data based on the credit overdue characterization value of each candidate feature data.

[0052] Among them, the bin risk control characterization value can be expressed by iv, iv = (bad sample / bad sample-good sample / good sample)*log((bad sample / bad sample) / (good sample / good sample)). Bad samples are credit data determined to be overdue in the group of chi-square bin data, and good samples are credit data determined to be unexpected in the group of chi-square bin data.

[0053] The credit overdue characterization value can be expressed by the bad debt rate, where bad debt rate = number of bad samples / total number of samples.

[0054] For example, historical credit data can be binned using chi-square binning, and boxes of similar bad samples can be combined to eventually present monotonicity. The corresponding iv is calculated and summed, and an initial screening is performed to select variable features with iv>0.05 or above. Subsequently, similar variable features are compared for effect, with priority given to those with good effects. Ideally, variables with iv>0.2 or above are screened to obtain target feature data.

[0055] S140. Use a random forest model to determine target risk control rules from a risk control rule set based on target feature data.

[0056] Among them, the random forest model uses decision trees as the basic classifier, and uses the bagging algorithm to train multiple decision tree models during training; during classification, the multiple decision trees involved make category predictions respectively, and then use the voting method to determine the category of the input data; the input data is classified into the category that receives the most votes. The application effect of the random forest model has been widely proven in theory and practice. It has excellent generalization ability, is not prone to overfitting, and has a small prediction error.

[0057] Random Sampling: Random Forest uses bootstrap sampling to create the training data set for each decision tree. This means that samples are randomly drawn from the original training data set with replacement to build each decision tree. Since the training data for each decision tree is obtained through sampling with replacement, each sample may appear multiple times in a tree or not at all.

[0058] Random feature selection: At each decision tree node split, random forests do not use all features for splitting, but randomly select a subset of features from all features to consider. This helps increase the diversity of each decision tree, making random forests more diverse.

[0059] Establishment of multiple decision trees: Random forests are composed of multiple decision trees, each of which is trained independently and has a certain degree of randomness. Each decision tree predicts the input data, and the final prediction result is obtained by combining (usually voting or averaging) the outputs of all decision trees.

[0060] Voting or averaging: For classification problems, random forests use a voting mechanism, that is, each decision tree votes for a category, and the final prediction result is the category with the most votes. For regression problems, random forests use an averaging mechanism, that is, each decision tree gives a prediction value, and the final prediction result is the average of these values.

[0061] Random forest improves the generalization and anti-overfitting capabilities of the model by introducing the ideas of randomness and ensemble learning. Using the random forest mining strategy, it is possible to quickly and deeply mine strategies that are difficult for humans to discover, greatly saving time and manpower, reducing the development cycle, and more accurately managing risks.

[0062] In some embodiments, the risk control rule set includes: a plurality of candidate risk control rules. The plurality of candidate risk control rules can be stored in a rule library so as to be effectively obtained when used.

[0063] A random forest model is used to determine the target risk control rule from the risk control rule set based on the target feature data, including: building a corresponding decision tree based on each candidate risk control rule; inputting the target feature data into the decision tree corresponding to each candidate risk control rule respectively, and obtaining the output result of the decision tree corresponding to each candidate risk control rule; and determining the target risk control rule from multiple candidate risk control rules based on the output result of the decision tree corresponding to each candidate risk control rule.

[0064] For example, if there are 6 decision trees preset in the random forest model and 18 candidate risk control rules, corresponding decision trees are established based on the first 6 candidate risk control rules, and then the target feature data is input into the random forest model to obtain the output result prediction value of each decision tree, and the candidate risk control rule corresponding to the maximum result prediction value is determined as a target risk control rule. By analogy, 3 target risk control rules can be determined from the 18 candidate risk control rules. This is convenient for effectively determining target risk control rules with higher decision-making power.

[0065] In some embodiments, the target risk control rule includes: multiple sub-risk control rules.

[0066] After using the random forest model to determine the target risk control rules from the risk control rule set based on the target feature data, it also includes:

[0067] Determine rule screening conditions based on historical credit data; based on the rule screening conditions, perform rule screening on multiple sub-risk control rules; based on the sub-risk control rules that meet the rule screening conditions, update the target risk control rules.

[0068] Among them, determining rule screening conditions based on historical credit data may include: determining credit characteristics with a sample number greater than a preset threshold from multiple credit data included in the historical credit data; determining the credit data corresponding to the credit characteristics with a sample number greater than the preset threshold as screening conditions, such as gender is male and age is 20-40 years old.

[0069] Based on the rule screening conditions, multiple sub-risk control rules are screened, which may include: matching the rule screening conditions with the rule features corresponding to each sub-risk control rule respectively, and if the two match, determining that the sub-risk control rule meets the rule screening conditions. In this way, the sub-risk control rule that meets the rule screening conditions is determined as a new target risk control rule.

[0070] Therefore, the sub-risk control rules that match the historical credit data are screened out through the rule screening conditions, so as to delete the target risk control rules and obtain more useful risk control rules.

[0071] S150. Perform data analysis on the target risk control rules to obtain credit risk control rules.

[0072] Among them, credit risk control rules are used to detect the credit risks of loan personnel.

[0073] In some embodiments, data analysis is performed on the target risk control rules to obtain the credit risk control rules, including:

[0074] Based on the credit overdue characterization value corresponding to each sub-risk control rule, a credit optimization rule is determined from multiple sub-risk control rules; the credit optimization rule is tested based on the target characteristic data to obtain the credit risk control rule.

[0075] For example, a rule with a high bad debt rate and a high hit rate is selected from multiple sub-risk control rules as a credit optimization rule, and the feature data used by the credit optimization rule is removed. The remaining target feature data is used to continue testing the rule to effectively derive the credit risk control rule. The credit risk control rule includes: multiple risk control rules, each of which can detect the credit risk of loan personnel.

[0076] Before testing the credit optimization rules based on the target feature data and obtaining the credit risk control rules, the target feature data can also be tested to obtain highly stable target feature data. For example, the historical credit data can be split into training sets and test sets according to time, and the rules can be verified on the training sets and test sets. If the stability meets the standard, the target feature data is determined to be useful and retained.

[0077] In addition, in this embodiment, after obtaining the credit risk control rules, the bad debt rate of the data that finally hits the rules and the remaining data is checked. If the bad debt rate of the data that hits the rules is much higher than that of the non-hit data and the number of loan rejections is within an acceptable range, it means that the effect is very good.

[0078] This embodiment integrates the advantages of multiple decision trees to achieve more accurate and robust judgment of user access, which not only improves the security of the system, but also has a strong ability to resist attacks and dynamic environments.

[0079] In this embodiment, historical credit data is obtained, and the historical credit data includes: pre-loan sample data, post-loan sample data and third-party sample data. The pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, the post-loan sample data is used to describe the credit data of the loan personnel after the loan, and the third-party sample data is used to describe the credit data stored in the third-party risk control platform; the historical credit data is subjected to chi-square binning processing to obtain multiple groups of chi-square binned data, and each group of card-square binned data includes at least one of the following: pre-loan sample data, post-loan sample data and third-party sample data; target feature data is determined from the multiple groups of chi-square binned data, and the target feature data is used to describe the credit data corresponding to the target credit feature in the historical credit data; a random forest model is used to determine the target risk control rule from the risk control rule set based on the target feature data; data analysis is performed on the target risk control rule to obtain the credit risk control rule, and the credit risk control rule is used to detect the credit risk of the loan personnel. In this way, by automatically binning the summarized historical credit data and using the random forest model to formulate rules for the target feature data obtained after binning, it is easy to obtain credit risk control rules for detecting the credit risk of loan personnel, thereby effectively improving risk control efficiency.

[0080] Figure 2 A structural schematic diagram of a risk control rule processing device based on random forest is provided for this embodiment. The risk control rule processing device based on random forest may include: an acquisition module 210, a processing module 220, a first determination module 230, a second determination module 240 and an analysis module 250.

[0081] Acquisition module 210 is used to acquire historical credit data, which includes: pre-loan sample data, post-loan sample data and third-party sample data. Pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, post-loan sample data is used to describe the credit data of the loan personnel after the loan, and third-party sample data is used to describe the credit data stored in the third-party risk control platform.

[0082] The processing module 220 is used to perform card-square binning processing on the historical credit data to obtain multiple groups of card-square binning data, each group of card-square binning data includes at least one of the following: pre-loan sample data, post-loan sample data and third-party sample data.

[0083] The first determination module 230 is used to determine target feature data from multiple groups of chi-square binned data, where the target feature data is used to describe the credit data corresponding to the target credit feature in the historical credit data.

[0084] The second determination module 240 is used to use a random forest model to determine a target risk control rule from a risk control rule set based on target feature data.

[0085] The analysis module 250 is used to perform data analysis on the target risk control rules to obtain credit risk control rules, which are used to detect the credit risk of loan personnel.

[0086] In this embodiment, optionally, the processing module 220 is specifically configured to:

[0087] Obtaining preset risk control indicators, which are used to describe overdue judgment indicators of credit data; obtaining risk control credit data from historical credit data based on the preset risk control indicators, wherein the risk control credit data includes: multiple sub-credit data; performing chi-square binning processing on the multiple sub-credit data based on risk control similarity characterization values ​​between the multiple sub-credit data, and obtaining multiple groups of chi-square binned data.

[0088] In this embodiment, optionally, the first determining module 230 is specifically configured to:

[0089] Obtain the bin risk control characterization value corresponding to each group of chi-square bin data; determine the chi-square bin data whose bin risk control characterization value is greater than the preset risk control characterization value as candidate feature data; determine the target feature data based on the credit overdue characterization value of each candidate feature data.

[0090] In this embodiment, optionally, the risk control rule set includes: multiple candidate risk control rules.

[0091] The second determining module 240 is specifically configured to:

[0092] Based on each candidate risk control rule, a corresponding decision tree is constructed; the target feature data is input into the decision tree corresponding to each candidate risk control rule respectively to obtain the output result of the decision tree corresponding to each candidate risk control rule; based on the output result of the decision tree corresponding to each candidate risk control rule, the target risk control rule is determined from multiple candidate risk control rules.

[0093] In this embodiment, optionally, the target risk control rule includes: multiple sub-risk control rules.

[0094] It also includes: a third determination module, a screening module and an updating module.

[0095] The third determination module is used to determine rule screening conditions based on historical credit data.

[0096] The screening module is used to screen multiple sub-risk control rules based on rule screening conditions.

[0097] The update module is used to update the target risk control rules based on the sub-risk control rules that meet the rule screening conditions.

[0098] In this embodiment, optionally, the analysis module 250 is specifically used to:

[0099] Based on the credit overdue characterization value corresponding to each sub-risk control rule, a credit optimization rule is determined from multiple sub-risk control rules; the credit optimization rule is tested based on the target characteristic data to obtain the credit risk control rule.

[0100] In this embodiment, optionally, the acquisition module 210 is further used to acquire credit data that meets a preset data type in the historical credit data. The preset data type includes: data missing type and data abnormality type.

[0101] The updating module is also used to perform data preprocessing on the credit data that meets the preset data type in the historical credit data so as to update the historical credit data.

[0102] Among them, data preprocessing includes: missing data filling and abnormal data replacement.

[0103] The risk control rule processing device based on random forest provided in the present disclosure can execute the above method embodiments. Its specific implementation principles and technical effects can be found in the above method embodiments, and the present disclosure will not repeat them here.

[0104] The present application also provides a computer device. Figure 3 , Figure 3 This is a basic structural block diagram of the computer device in this embodiment.

[0105] The computer device includes a memory 310 and a processor 320 that are connected to each other through a system bus. It should be noted that the figure only shows a computer device with a memory 310 and a processor 320, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0106] Computer devices can be computing devices such as desktop computers, notebooks, PDAs, and cloud servers. Computer devices can interact with users through keyboards, mice, remote controls, touch pads, or voice control devices.

[0107] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes a non-volatile memory or a volatile memory, for example, a flash memory, a hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc., and the RAM may include a static RAM or a dynamic RAM. In some embodiments, the memory 310 may be an internal storage unit of a computer device, for example, a hard disk or a memory of the computer device. In other embodiments, the memory 310 may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, or a flash card (FlashCard) equipped on the computer device. Of course, the memory 310 may also include both the internal storage unit of the computer device and its external storage device. In this embodiment, the memory 310 is generally used to store the operating system and various application software installed on the computer device, such as the program code of the above method. In addition, the memory 310 may also be used to temporarily store various data that have been output or are to be output.

[0108] The processor 320 is generally used to perform the overall operation of the computer device. In this embodiment, the memory 310 is used to store program codes or instructions, the program code includes computer operation instructions, and the processor 320 is used to execute the program codes or instructions stored in the memory 310 or process data, such as running the program code of the above method.

[0109] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus system can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

[0110] Another embodiment of the present application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads a computer-readable program code stored in the computer-readable medium, so that the processor can execute the functional actions specified in each step or a combination of steps in the above method; and generate a device for implementing the functional actions specified in each block or a combination of blocks in the block diagram.

[0111] Computer-readable media include but are not limited to electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any appropriate combination of the foregoing, the memory is used to store program codes or instructions, the program codes include computer operating instructions, and the processor is used to execute the program codes or instructions of the above methods stored in the memory.

[0112] For the definitions of memory and processor, please refer to the description of the aforementioned computer device embodiment and will not be repeated here.

[0113] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0114] Each functional unit or module in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0115] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) or a processor (processor) to perform all or part of the steps of each embodiment method of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), disk or optical disk and other media that can store program code.

[0116] In the claims, any reference symbols placed between brackets shall not be construed as limiting the claims. The word "comprising" described in the present application does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present application can be implemented with the aid of hardware comprising several different elements and with the aid of a suitably programmed computer. In a unit claim that lists a number of devices, several units of these devices may be embodied by the same hardware item. The use of first, second, and third, etc. does not indicate any order, and these words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

[0117] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A risk control rule processing method based on random forest, characterized in that: include: Obtaining historical credit data, the historical credit data includes: pre-loan sample data, post-loan sample data and third-party sample data, the pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, the post-loan sample data is used to describe the credit data of the loan personnel after the loan, and the third-party sample data is used to describe the credit data stored in the third-party risk control platform; Performing card-square binning processing on the historical credit data to obtain multiple groups of card-square binning data, each group of the card-square binning data including at least one of the following: the pre-loan sample data, the post-loan sample data, and the third-party sample data; Determining target feature data from the plurality of groups of the chi-square binned data, the target feature data being used to describe the credit data corresponding to the target credit feature in the historical credit data; Using a random forest model, a target risk control rule is determined from a risk control rule set based on the target feature data; Data analysis is performed on the target risk control rules to obtain credit risk control rules, and the credit risk control rules are used to detect the credit risk of loan personnel.

2. The method according to claim 1, characterized in that The card-square binning process is performed on the historical credit data to obtain multiple groups of card-square binning data, including: Obtaining a preset risk control indicator, wherein the preset risk control indicator is used to describe an overdue determination indicator of credit data; Based on the preset risk control indicator, obtaining risk control credit data from the historical credit data, wherein the risk control credit data includes: a plurality of sub-credit data; Based on the risk control similarity characterization values ​​between the plurality of sub-credit data, the plurality of sub-credit data are subjected to chi-square binning processing to obtain a plurality of groups of chi-square binned data.

3. The method according to claim 1, characterized in that The step of determining target feature data from the plurality of groups of chi-square binned data comprises: Obtain the bin risk control representation value corresponding to each group of the chi-square bin data; Determine the chi-square binned data whose binned risk control characterization value is greater than the preset risk control characterization value as candidate feature data; The target feature data is determined based on the credit overdue characterization value of each of the candidate feature data.

4. The method according to claim 1, characterized in that: The risk control rule set includes: a plurality of candidate risk control rules; The adopting of the random forest model to determine the target risk control rule from the risk control rule set based on the target feature data includes: Based on each of the candidate risk control rules, construct a corresponding decision tree; Inputting the target feature data into the decision tree corresponding to each of the candidate risk control rules respectively, and obtaining the output result of the decision tree corresponding to each of the candidate risk control rules; Based on the output results of the decision tree corresponding to each of the candidate risk control rules, the target risk control rule is determined from multiple candidate risk control rules.

5. The method according to claim 1, characterized in that The target risk control rule includes: a plurality of sub-risk control rules; After the random forest model is used to determine the target risk control rule from the risk control rule set based on the target feature data, the method further includes: Determining rule screening conditions based on the historical credit data; Based on the rule screening condition, the plurality of sub-risk control rules are screened; Based on the sub-risk control rules that meet the rule screening conditions, the target risk control rules are updated.

6. The method according to claim 5, characterized in that The performing of data analysis on the target risk control rule to obtain the credit risk control rule includes: Determining a credit optimization rule from a plurality of the sub-risk control rules based on the credit overdue characterization value corresponding to each of the sub-risk control rules; The credit optimization rule is tested based on the target feature data to obtain the credit risk control rule.

7. The method according to claim 1, characterized in that Before performing card-square binning processing on the historical credit data to obtain multiple groups of card-square binning data, the method further includes: Acquire the credit data that meets the preset data type in the historical credit data, wherein the preset data type includes: data missing type and data abnormality type; Performing data preprocessing on the credit data that meets the preset data type in the historical credit data to update the historical credit data; Wherein, the data preprocessing includes: missing data filling and abnormal data replacement.

8. A risk control rule processing device based on random forest, characterized in that: include: An acquisition module is used to acquire historical credit data, wherein the historical credit data includes: pre-loan sample data, post-loan sample data and third-party sample data. The pre-loan sample data is used to describe the identity data and behavior data of the loan personnel before the loan, the post-loan sample data is used to describe the credit data of the loan personnel after the loan, and the third-party sample data is used to describe the credit data stored in the third-party risk control platform; A processing module, configured to perform card-square binning processing on the historical credit data to obtain multiple groups of card-square binning data, each group of the card-square binning data including at least one of the following: the pre-loan sample data, the post-loan sample data, and the third-party sample data; A first determination module is used to determine target feature data from multiple groups of the chi-square binned data, wherein the target feature data is used to describe the credit data corresponding to the target credit feature in the historical credit data; A second determination module is used to determine a target risk control rule from a risk control rule set based on the target feature data by using a random forest model; The analysis module is used to perform data analysis on the target risk control rules to obtain credit risk control rules, and the credit risk control rules are used to detect the credit risk of loan personnel.

9. A computer device, characterized in that: It includes a memory and a processor, the memory stores a computer program, and the processor implements the risk control rule processing method based on random forest as described in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the risk control rule processing method based on random forest is implemented as described in any one of claims 1 to 7.

Images