WO2019061989A1 - 贷款风险控制方法、电子装置及可读存储介质 - Google Patents

贷款风险控制方法、电子装置及可读存储介质 Download PDF

Info

Publication number
WO2019061989A1
WO2019061989A1 PCT/CN2018/076139 CN2018076139W WO2019061989A1 WO 2019061989 A1 WO2019061989 A1 WO 2019061989A1 CN 2018076139 W CN2018076139 W CN 2018076139W WO 2019061989 A1 WO2019061989 A1 WO 2019061989A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
loan risk
loan
preset application
preset
Prior art date
Application number
PCT/CN2018/076139
Other languages
English (en)
French (fr)
Inventor
金新
王建明
肖京
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2019061989A1 publication Critical patent/WO2019061989A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/03—Credit; Loans; Processing thereof
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines

Definitions

  • the present application relates to the field of computer technologies, and in particular, to a loan risk control method, an electronic device, and a readable storage medium.
  • the purpose of the application is to provide a loan risk control method, an electronic device and a readable storage medium, aiming at reducing loan risk.
  • a first aspect of the present application provides an electronic device, where the electronic device includes a memory and a processor, where the memory stores a loan risk control system operable on the processor, and the loan risk
  • the control system implements the following steps when executed by the processor:
  • a user model data training based on different loan risk levels obtains a hierarchical model of the loan risk level, and uses the hierarchical model to classify users on the preset application into different loan risk levels;
  • the second aspect of the present application further provides a loan risk control method, where the loan risk control method includes:
  • Step 1 Associate the seed users of different loan risk levels with the pre-established user images to obtain user portrait data of different loan risk levels
  • Step 2 A user model data training based on different loan risk levels obtains a hierarchical model of the loan risk level, and uses the hierarchical model to classify users on the preset application into different loan risk levels;
  • Step 3 According to different loan risk levels corresponding to the user on the preset application, and performing risk scoring on the user in the preset application according to the preset scoring rule, only the default loan entry is displayed to the user whose score reaches the preset requirement.
  • a third aspect of the present application further provides a computer readable storage medium storing a loan risk control system, the loan risk control system being executable by at least one processor And causing the at least one processor to perform the steps of the loan risk control method as described above.
  • the loan risk control method, system and readable storage medium proposed by the present application by associating seed users of different loan risk levels with user portraits, obtain user portrait data of different loan risk levels, and user images based on different loan risk levels
  • the data training obtains a hierarchical model of the loan risk level, and uses the hierarchical model to classify the users on the preset application into different loan risk levels; according to the different loan risk levels corresponding to the users, the risk score is scored, and only the score reaches the preset requirement.
  • the user presents a preset loan entry.
  • the loan risk score can be scored for the user on the preset application, only the loan entry on the preset application is displayed for the user whose score meets the requirements, and the loan success rate of promoting the loan business on the preset application of the diversion is improved.
  • inferior customers can also apply for loans through the loan entrance at will, which reduces the risk of the loan company and improves the conversion rate of the loan users.
  • FIG. 1 is a schematic diagram of an optional application environment of each embodiment of the present application.
  • FIG. 2 is a schematic diagram of a hardware architecture of an embodiment of the electronic device of FIG. 1;
  • FIG. 3 is a schematic flowchart of an embodiment of a loan risk control method according to the present application.
  • first, second and the like in the present application are for the purpose of description only, and are not to be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. .
  • features defining “first” and “second” may include at least one of the features, either explicitly or implicitly.
  • the technical solutions between the various embodiments may be combined with each other, but must be based on the realization of those skilled in the art, and when the combination of the technical solutions is contradictory or impossible to implement, it should be considered that the combination of the technical solutions does not exist. Nor is it within the scope of protection required by this application.
  • FIG. 1 is a schematic diagram of an optional application environment of each embodiment of the present application.
  • the present application is applicable to an application environment including, but not limited to, an electronic device 1, a terminal device 2, and a network 3.
  • the electronic device 1 is an apparatus capable of automatically performing numerical calculation and/or information processing in accordance with an instruction set or stored in advance.
  • the electronic device 1 may be a computer, a single network server, a server group composed of a plurality of network servers, or a cloud-based cloud composed of a large number of hosts or network servers, wherein the cloud computing is a type of distributed computing, A super virtual computer consisting of a loosely coupled set of computers.
  • the terminal device 2 includes, but is not limited to, any electronic product that can interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device, for example, a personal computer, a tablet computer, a smart phone, or an individual.
  • Digital Assistant (PDA) game console, Internet Protocol Television (IPTV), smart wearable device, etc.
  • the network 3 may be an intranet, an Internet, a global mobile communication system (GSM), and a Wideband Code Division Multiple Access (WCDMA). Wireless or wired networks such as 4G networks, 5G networks, Bluetooth, Wi-Fi, etc.
  • the electronic device 1 is communicatively connected to one or more of the terminal devices 2 via the network 3 respectively.
  • FIG. 2 is a schematic diagram of an optional hardware architecture of the electronic device 1 of FIG. 1.
  • the electronic device 1 may include, but is not limited to, a memory 11 and a processor 12 that are communicably connected to each other through a system bus. Network interface 13. It is to be noted that FIG. 2 only shows the electronic device 1 having the components 11-13, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead.
  • the memory 11 includes at least one type of readable storage medium including a flash memory, a hard disk, a multimedia card, a card type memory (eg, SD or DX memory, etc.), and a random access memory (RAM). , static random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disk, optical disk, and the like.
  • the memory 11 may be an internal storage unit of the electronic device 1, such as a hard disk or a memory of the electronic device 1.
  • the memory 11 may also be an external storage device of the electronic device 1, such as a plug-in hard disk equipped on the electronic device 1, a smart memory card (SMC), and a secure digital device. (Secure Digital, SD) card, flash card, etc.
  • SMC smart memory card
  • SD Secure Digital
  • the memory 11 can also include both the internal storage unit of the electronic device 1 and its external storage device.
  • the memory 11 is generally used to store an operating system installed in the electronic device 1 and various types of application software, such as program codes of the loan risk control system 10, and the like. Further, the memory 11 can also be used to temporarily store various types of data that have been output or are to be output.
  • the processor 12 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments.
  • the processor 12 is typically used to control the overall operation of the electronic device 1, such as performing control and processing related to data interaction or communication with the terminal device 2.
  • the processor 12 is configured to run program code or process data stored in the memory 11, such as running the loan risk control system 10 and the like.
  • the network interface 13 may comprise a wireless network interface or a wired network interface, which is typically used to establish a communication connection between the electronic device 1 and other electronic devices.
  • the network interface 13 is mainly used to connect the electronic device 1 to one or more of the terminal devices 2 through the network 3, in the electronic device 1 and one or more of the terminals. A data transmission channel and a communication connection are established between the devices 2.
  • the loan risk control system 10 includes at least one computer readable instruction stored in the memory 11, the at least one computer readable instruction being executable by the processor 12 to implement various embodiments of the present application.
  • step S1 the seed users of different loan risk levels are associated with the pre-established user images to obtain user portrait data of different loan risk levels.
  • a batch of seed users is first obtained, and the seed users are classified into loan risk levels.
  • a group of representative and universal seed users can be selected from the user database, and the seed users can be classified according to experience, such as high risk, medium risk, low risk and other loan risk levels. .
  • a pre-established user portrait is obtained, and selected seed users of different loan risk levels are associated with pre-established user portraits to obtain user portrait data of different loan risk levels.
  • the user portrait is a target user model based on a series of real data, and is a tagged user model abstracted based on information such as user social attributes, living habits and consumer behavior.
  • the core task of building user portraits is to label users with "high-profile" features that are analyzed by user information.
  • the preset user portrait in this embodiment may directly call the established user portrait related to the loan personnel, or may be through various data sources (such as a loan website database, QQ, Weibo, WeChat, Snowball, Oriental).
  • the user portraits created include various attribute tags, such as the user's social attribute tags and financial attribute tags, such as age, family status, income, occupation, spending habits, whether there is a loan record Have you ever done a credit card, etc.
  • the seed users of different loan risk levels are associated with corresponding user images containing attribute tags, that is, each attribute in the seed user data, such as age, gender, preference, income, etc., is matched with various attribute tags corresponding to the user portrait.
  • the seed user data with high matching degree is associated with the user image, and all the attribute tags in the user portrait are assigned to the seed user data associated with the user image data to form the associated user image data.
  • the seed user data which has only basic attributes, also has various attribute tags in the user portrait, such as various social attribute tags and financial attribute tags, so as to facilitate subsequent modeling and more accurate introduction of the promotion loan business.
  • the rating of the loan risk level is calculated by the user on the streaming application.
  • Step S2 The user model data of different loan risk levels is trained to obtain a classification model of the loan risk level, and the user of the preset application is classified by the classification model, and is divided into different loan risk levels.
  • the user portrait data of the different loan risk levels after the association is obtained, and the hierarchical model of the loan risk level can be trained based on the user portrait data of different loan risk levels.
  • one-versus-rest OVR SVMs
  • the sample of a certain category is a user of a certain loan risk level.
  • the image data is classified into one class, and the other remaining samples are classified into another class, so that k samples of the class structure construct k support vector machine SVMs, and the unknown samples are classified into the class with the largest classification function value.
  • one-versus-one may be used to train the hierarchical model in either of two types of samples, ie, two loan risks. Design an SVM between the user image data of the level, so the k categories need to design k(k-1)/2 SVMs. When classifying an unknown sample, the category with the most votes is the unknown sample. Category.
  • a hierarchical model can be assumed, and the loan risk level can be assumed to include four levels a. , b, c, d, then the training process of the hierarchical model is as follows:
  • the user portrait data vector corresponding to the loan risk level a is extracted as the positive set, and the user portrait data vector corresponding to the loan risk levels b, c, and d is used as the negative set to obtain the first training set;
  • the user portrait data vector corresponding to the loan risk level b is used as a positive set, and the user portrait data vector corresponding to the loan risk levels a, c, and d is extracted as a negative set to obtain a second training set; and the corresponding loan risk level c is extracted.
  • the user portrait data vector is used as a positive set, and the user portrait data vector corresponding to the loan risk levels a, b, and d is used as a negative set to obtain a third training set; and the user portrait data vector corresponding to the loan risk level d is extracted as a positive set.
  • the user portrait data vector corresponding to the loan risk levels a, b, and c is used as a negative set to obtain a fourth training set; the four training sets are separately trained to obtain four support vector machine SVMs as the classification of the loan risk level. model.
  • the training sets of the first training set, the second training set, the third training set, and the fourth training set are respectively trained, and then the training result files of the four support vector machine SVMs are obtained, and in the subsequent testing, the corresponding The test vectors are tested using the four training result files, respectively. Finally, each test has a result, and the final result is the largest of the four values.
  • Step S3 According to different loan risk levels corresponding to the user on the preset application, and performing risk scoring on the user on the preset application according to the preset scoring rule, only the default loan entry is displayed to the user whose score reaches the preset requirement.
  • the risk scores can be performed for each user. For example, if different loan risk levels are pre-set to include four levels: high, medium to upper, medium to low, and low, then among users, users with high, medium, and medium-lower loan risk levels can be set. If the risk score is low, only the default loan entry will be presented to users with low risk scores. In the risk scoring, the loan risk score can be comprehensively combined with the user's loan risk level based on pre-set business rules, user categories, user tags, and the user's current location address.
  • the user who meets the preset requirement displays the default loan entry, and the high-risk user whose score does not meet the preset requirement does not display the default loan entry, so as to prevent the high-risk user from handling the loan business through the preset loan entrance.
  • the preset loan interface may be displayed at a preset location such as a bottom position, a top end, and the like at a key position of the APP interface according to the device number when the terminal device of the user whose rating reaches the preset requirement displays the preset application interface. To facilitate the user to guide the credit to reach the preset requirements.
  • the present embodiment associates seed users of different loan risk levels with user portraits to obtain user portrait data of different loan risk levels, and trains loan risk levels based on user portrait data of different loan risk levels.
  • the grading model is used to classify users on the preset application into different loan risk levels; perform risk grading according to different loan risk levels corresponding to the user, and only display preset loan entries to users whose scores reach the preset requirements . Since the loan risk score can be scored for the user on the preset application, only the loan entry on the preset application is displayed for the user whose score meets the requirements, and the loan success rate of promoting the loan business on the preset application of the diversion is improved. To avoid users who do not meet the requirements, that is, inferior customers can also apply for loans through the loan entrance at will, which reduces the risk of the loan company and improves the conversion rate of the loan users.
  • the method further includes:
  • Obtaining behavior data of the user on the preset application, and clustering the users on the preset application into different behavior categories according to the behavior data, and calculating a risk score M of the user on the preset application is:
  • M1 is a predetermined score base corresponding to different loan risk levels
  • x1 is a weight coefficient corresponding to the loan risk level
  • M2 is a preset score base corresponding to different behavior categories
  • y1 is a weight coefficient corresponding to the behavior category.
  • the behavior data of the user on the preset application may be comprehensively considered, and the users on the preset application are clustered into different behavior categories according to the behavior data, for example, may be preset according to different users. Consumer behavior data on the application will be clustered into categories such as regular consumption, occasional consumption, and basic non-consumption, or in categories such as large-volume consumption and customary small-value consumption.
  • the scoring base M1 corresponding to different loan risk levels and its corresponding weight coefficient x1, and the scoring base M2 corresponding to different behavior categories and their corresponding weight coefficients y1 are pre-set, and the user's risk score M is comprehensively calculated. .
  • the scores of different loan risk levels such as high, medium upper, middle lower, and low correspond to 1, 0.8, 0.6, and 0.4, respectively, and the weight coefficient x1 corresponding to the loan risk level is 0.4.
  • the scores of different behavior categories such as frequent consumption, occasional consumption, and basic non-consumption categories are 0.5, 0.7, and 0.9, respectively, and the weight coefficient corresponding to the behavior category is 0.6.
  • the risk scores of each user on the preset application are calculated in turn. The higher the score, the higher the risk of the loan. Therefore, only the users whose scores reach the preset requirements, such as below a certain threshold, are presented with a preset loan entry.
  • the user's loan risk level and the user's behavior data on the preset application can be comprehensively evaluated, and the user's loan risk can be more accurately and comprehensively evaluated, so that the loan can be more accurately performed. risk control.
  • the method further includes:
  • M3 is the default score base corresponding to different loan risk levels
  • x3 is the weight coefficient corresponding to the loan risk level
  • M4 is the default score base corresponding to different labels
  • y4 is the weight coefficient corresponding to the label.
  • the positioning data of the user on the preset application program may be comprehensively considered, and the corresponding preset label is set for the user according to the positioning data and the user portrait data corresponding to the loan risk level of the user.
  • the corresponding preset label can be set for the user as “ If the location address is an ordinary shopping mall, and the income attribute label in the user portrait data corresponding to the user is a middle income group, the corresponding preset label can be set for the user as “medium consumer group”, etc. .
  • the scoring base M3 corresponding to different loan risk levels and the corresponding weight coefficient x3, and the scoring base M4 corresponding to different tags and the corresponding weight coefficient y4 are preset, and the user's risk score M is comprehensively calculated.
  • the scores of different loan risk levels such as high, medium to upper, medium to low, and low may be set to 1, 0.8, 0.6, and 0.4, respectively, and the weight coefficient x1 corresponding to the loan risk level is 0.3.
  • the scores of different labels such as high-consumer population, medium-consumer group, and low-consumption group are 0.3, 0.5, and 0.9, respectively, and the weight coefficient corresponding to the behavior category is 0.7.
  • the risk scores of each user on the preset application are calculated in turn. The higher the score, the higher the risk of the loan. Therefore, only the users whose scores reach the preset requirements, such as below a certain threshold, are presented with a preset loan entry.
  • the user's loan risk level and the user's positioning data on the preset application can be comprehensively evaluated, and the user's loan risk can be more accurately and comprehensively evaluated to more accurately perform the loan. risk control.
  • FIG. 3 is a schematic flowchart of an embodiment of a loan risk control method according to the present application.
  • the loan risk control method includes the following steps:
  • step S10 the seed users of different loan risk levels are associated with the pre-established user images to obtain user portrait data of different loan risk levels.
  • a batch of seed users is first obtained, and the seed users are classified into loan risk levels.
  • a group of representative and universal seed users can be selected from the user database, and the seed users can be classified according to experience, such as high risk, medium risk, low risk and other loan risk levels. .
  • a pre-established user portrait is obtained, and selected seed users of different loan risk levels are associated with pre-established user portraits to obtain user portrait data of different loan risk levels.
  • the user portrait is a target user model based on a series of real data, and is a tagged user model abstracted based on information such as user social attributes, living habits and consumer behavior.
  • the core task of building user portraits is to label users with "high-profile" features that are analyzed by user information.
  • the preset user portrait in this embodiment may directly call the established user portrait related to the loan personnel, or may be through various data sources (such as a loan website database, QQ, Weibo, WeChat, Snowball, Oriental).
  • the user portraits created include various attribute tags, such as the user's social attribute tags and financial attribute tags, such as age, family status, income, occupation, spending habits, whether there is a loan record Have you ever done a credit card, etc.
  • the seed users of different loan risk levels are associated with the corresponding user images containing the attribute tags, that is, the attributes of the seed user data, such as age, gender, preference, income, etc., are matched with corresponding attribute tags in the user portrait.
  • the seed user data with high matching degree is associated with the user image, and all the attribute tags in the user portrait are assigned to the seed user data associated with the user image data to form the associated user image data.
  • the seed user data which has only basic attributes, also has various attribute tags in the user portrait, such as various social attribute tags and financial attribute tags, so as to facilitate subsequent modeling and more accurate introduction of the promotion loan business.
  • the rating of the loan risk level is calculated by the user on the streaming application.
  • Step S20 The user model data of different loan risk levels is trained to obtain a classification model of the loan risk level, and the user of the preset application is classified by the classification model, and is divided into different loan risk levels.
  • the user portrait data of the different loan risk levels after the association is obtained, and the hierarchical model of the loan risk level can be trained based on the user portrait data of different loan risk levels.
  • one-versus-rest OVR SVMs
  • the sample of a certain category is a user of a certain loan risk level.
  • the image data is classified into one class, and the other remaining samples are classified into another class, so that k samples of the class structure construct k support vector machine SVMs, and the unknown samples are classified into the class with the largest classification function value.
  • one-versus-one may be used to train the hierarchical model in either of two types of samples, ie, two loan risks. Design an SVM between the user image data of the level, so the k categories need to design k(k-1)/2 SVMs. When classifying an unknown sample, the category with the most votes is the unknown sample. Category.
  • a hierarchical risk model can be assumed, and the loan risk level can be assumed to include four levels a. , b, c, d, then the training process of the hierarchical model is as follows:
  • the user portrait data vector corresponding to the loan risk level a is extracted as the positive set, and the user portrait data vector corresponding to the loan risk levels b, c, and d is used as the negative set to obtain the first training set;
  • the user portrait data vector corresponding to the loan risk level b is used as a positive set, and the user portrait data vector corresponding to the loan risk levels a, c, and d is extracted as a negative set to obtain a second training set; and the corresponding loan risk level c is extracted.
  • the user portrait data vector is used as a positive set, and the user portrait data vector corresponding to the loan risk levels a, b, and d is used as a negative set to obtain a third training set; and the user portrait data vector corresponding to the loan risk level d is extracted as a positive set.
  • the user portrait data vector corresponding to the loan risk levels a, b, and c is used as a negative set to obtain a fourth training set; the four training sets are separately trained to obtain four support vector machine SVMs as the classification of the loan risk level. model.
  • the training sets of the first training set, the second training set, the third training set, and the fourth training set are respectively trained, and then the training result files of the four support vector machine SVMs are obtained, and in the subsequent testing, the corresponding The test vectors are tested using the four training result files, respectively. Finally, each test has a result, and the final result is the largest of the four values.
  • Step S30 According to different loan risk levels corresponding to the user on the preset application, and performing risk scoring on the user in the preset application according to the preset scoring rule, only the default loan entry is displayed to the user whose score reaches the preset requirement.
  • the risk scores can be performed for each user. For example, if different loan risk levels are pre-set to include four levels: high, medium to upper, medium to low, and low, then among users, users with high, medium, and medium-lower loan risk levels can be set. If the risk score is low, only the default loan entry will be presented to users with low risk scores. In the risk scoring, the loan risk score can be comprehensively combined with the user's loan risk level based on pre-set business rules, user categories, user tags, and the user's current location address.
  • the user who meets the preset requirement displays the default loan entry, and the high-risk user whose score does not meet the preset requirement does not display the default loan entry, so as to prevent the high-risk user from handling the loan business through the preset loan entrance.
  • the preset loan interface may be displayed at a preset location such as a bottom position, a top end, and the like at a key position of the APP interface according to the device number when the terminal device of the user whose rating reaches the preset requirement displays the preset application interface. To facilitate the user to guide the credit to reach the preset requirements.
  • the present embodiment associates seed users of different loan risk levels with user portraits to obtain user portrait data of different loan risk levels, and trains loan risk levels based on user portrait data of different loan risk levels.
  • the grading model is used to classify users on the preset application into different loan risk levels; perform risk grading according to different loan risk levels corresponding to the user, and only display preset loan entries to users whose scores reach the preset requirements . Since the loan risk score can be scored for the user on the preset application, only the loan entry on the preset application is displayed for the user whose score meets the requirements, and the loan success rate of promoting the loan business on the preset application of the diversion is improved. To avoid users who do not meet the requirements, that is, inferior customers can also apply for loans through the loan entrance at will, which reduces the risk of the loan company and improves the conversion rate of the loan users.
  • step S30 further includes:
  • Obtaining behavior data of the user on the preset application, and clustering the users on the preset application into different behavior categories according to the behavior data, and calculating a risk score M of the user on the preset application is:
  • M1 is a predetermined score base corresponding to different loan risk levels
  • x1 is a weight coefficient corresponding to the loan risk level
  • M2 is a preset score base corresponding to different behavior categories
  • y1 is a weight coefficient corresponding to the behavior category.
  • the behavior data of the user on the preset application may be comprehensively considered, and the users on the preset application are clustered into different behavior categories according to the behavior data, for example, may be preset according to different users. Consumer behavior data on the application will be clustered into categories such as regular consumption, occasional consumption, and basic non-consumption, or in categories such as large-volume consumption and customary small-value consumption.
  • the scoring base M1 corresponding to different loan risk levels and its corresponding weight coefficient x1, and the scoring base M2 corresponding to different behavior categories and their corresponding weight coefficients y1 are pre-set, and the user's risk score M is comprehensively calculated. .
  • the scores of different loan risk levels such as high, medium upper, middle lower, and low correspond to 1, 0.8, 0.6, and 0.4, respectively, and the weight coefficient x1 corresponding to the loan risk level is 0.4.
  • the scores of different behavior categories such as frequent consumption, occasional consumption, and basic non-consumption categories are 0.5, 0.7, and 0.9, respectively, and the weight coefficient corresponding to the behavior category is 0.6.
  • the risk scores of each user on the preset application are calculated in turn. The higher the score, the higher the risk of the loan. Therefore, only the users whose scores reach the preset requirements, such as below a certain threshold, are presented with a preset loan entry.
  • the user's loan risk level and the user's behavior data on the preset application can be comprehensively evaluated, and the user's loan risk can be more accurately and comprehensively evaluated, so that the loan can be more accurately performed. risk control.
  • step S3 further includes:
  • M3 is the default score base corresponding to different loan risk levels
  • x3 is the weight coefficient corresponding to the loan risk level
  • M4 is the default score base corresponding to different labels
  • y4 is the weight coefficient corresponding to the label.
  • the positioning data of the user on the preset application program may be comprehensively considered, and the corresponding preset label is set for the user according to the positioning data and the user portrait data corresponding to the loan risk level of the user.
  • the corresponding preset label can be set for the user as “ If the location address is an ordinary shopping mall, and the income attribute label in the user portrait data corresponding to the user is a middle income group, the corresponding preset label can be set for the user as “medium consumer group”, etc. .
  • the scoring base M3 corresponding to different loan risk levels and the corresponding weight coefficient x3, and the scoring base M4 corresponding to different tags and the corresponding weight coefficient y4 are preset, and the user's risk score M is comprehensively calculated.
  • the scores of different loan risk levels such as high, medium to upper, medium to low, and low may be set to 1, 0.8, 0.6, and 0.4, respectively, and the weight coefficient x1 corresponding to the loan risk level is 0.3.
  • the scores of different labels such as high-consumer population, medium-consumer group, and low-consumption group are 0.3, 0.5, and 0.9, respectively, and the weight coefficient corresponding to the behavior category is 0.7.
  • the risk scores of each user on the preset application are calculated in turn. The higher the score, the higher the risk of the loan. Therefore, only the users whose scores reach the preset requirements, such as below a certain threshold, are presented with a preset loan entry.
  • the user's loan risk level and the user's positioning data on the preset application can be comprehensively evaluated, and the user's loan risk can be more accurately and comprehensively evaluated to more accurately perform the loan. risk control.
  • the present application also provides a computer readable storage medium storing a loan risk control system, the loan risk control system being executable by at least one processor to cause the at least one processor.
  • a computer readable storage medium storing a loan risk control system, the loan risk control system being executable by at least one processor to cause the at least one processor.
  • the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and can also be implemented by hardware, but in many cases, the former is A better implementation.
  • the technical solution of the present application which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk,
  • the optical disc includes a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the methods described in various embodiments of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Artificial Intelligence (AREA)
  • Strategic Management (AREA)
  • Marketing (AREA)
  • Economics (AREA)
  • Development Economics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Technology Law (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)

Abstract

本申请涉及一种贷款风险控制方法、电子装置及可读存储介质，该方法包括：将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据；基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进行分级，分为不同贷款风险等级；根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。本申请降低了贷款公司的风险，提升贷款用户的转化率。

Description

贷款风险控制方法、电子装置及可读存储介质
优先权申明
本申请基于巴黎公约申明享有2017年9月30日递交的申请号为CN 201710916484.8、名称为“贷款风险控制方法、电子装置及可读存储介质”中国专利申请的优先权，该中国专利申请的整体内容以参考的方式结合在本申请中。
技术领域
本申请涉及计算机技术领域，尤其涉及一种贷款风险控制方法、电子装置及可读存储介质。
背景技术
当前各贷款公司在导流APP上推广贷款业务时，主要是采用广撒网的方式，即针对所有用户展现贷款入口，并没有考虑到用户贷款的风险控制问题，劣质客户也能随意通过贷款入口办理贷款，加大了贷款风险，降低了贷款成功率。
发明内容
本申请的目的在于提供一种贷款风险控制方法、电子装置及可读存储介质，旨在降低贷款风险。
为实现上述目的，本申请第一方面提供一种电子装置，所述电子装置包括存储器、处理器，所述存储器上存储有可在所述处理器上运行的贷款风险控制系统，所述贷款风险控制系统被所述处理器执行时实现如下步骤：
A、将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据；
B、基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进行分级，分为不同贷款风险等级；
C、根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。
此外，为实现上述目的，本申请第二方面还提供一种贷款风险控制方法，所述贷款风险控制方法包括：
步骤一、将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据；
步骤二、基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进 行分级，分为不同贷款风险等级；
步骤三、根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。
进一步地，为实现上述目的，本申请第三方面还提供一种计算机可读存储介质，所述计算机可读存储介质存储有贷款风险控制系统，所述贷款风险控制系统可被至少一个处理器执行，以使所述至少一个处理器执行如上述的贷款风险控制方法的步骤。
本申请提出的贷款风险控制方法、系统及可读存储介质，通过将不同贷款风险等级的种子用户与用户画像进行关联，得到不同贷款风险等级的用户画像数据，并基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，利用所述分级模型将预设应用程序上的用户分为不同贷款风险等级；根据用户对应的不同贷款风险等级进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。由于能对预设应用程序上的用户进行贷款风险评分，仅对评分符合要求的用户展现预设应用程序上的贷款入口，提高了在导流的预设应用程序上推广贷款业务的贷款成功率，避免评分不符合要求的用户即劣质客户也能随意通过贷款入口办理贷款，降低了贷款公司的风险，提升贷款用户的转化率。
附图说明
图1为本申请各个实施例一可选的应用环境示意图；
图2是图1中电子装置一实施例的硬件架构的示意图；
图3为本申请贷款风险控制方法一实施例的流程示意图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
需要说明的是，在本申请中涉及“第一”、“第二”等的描述仅用于描述目的，而不能理解为指示或暗示其相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。另外，各个实施例之间的技术方案可以相互结合，但是必须是以本领域普通技术人员能够实现为基础，当技术方案的结合出现相互矛盾或无法实现时应当认为这种 技术方案的结合不存在，也不在本申请要求的保护范围之内。
参阅图1，是本申请各个实施例一可选的应用环境示意图。
在本实施例中，本申请可应用于包括，但不仅限于，电子装置1、终端设备2、网络3的应用环境中。其中，电子装置1是一种能够按照事先设定或者存储的指令，自动进行数值计算和/或信息处理的设备。电子装置1可以是计算机、也可以是单个网络服务器、多个网络服务器组成的服务器组或者基于云计算的由大量主机或者网络服务器构成的云，其中云计算是分布式计算的一种，由一群松散耦合的计算机集组成的一个超级虚拟计算机。
终端设备2包括，但不限于，任何一种可与用户通过键盘、鼠标、遥控器、触摸板或者声控设备等方式进行人机交互的电子产品，例如，个人计算机、平板电脑、智能手机、个人数字助理(Personal Digital Assistant，PDA)，游戏机、交互式网络电视(Internet Protocol Television,IPTV)、智能式穿戴式设备等。
所述网络3可以是企业内部网(Intranet)、互联网(Internet)、全球移动通讯系统(Global S贷款风险控制stem of Mobile communication，GSM)、宽带码分多址(Wideband Code Division Multiple Access，WCDMA)、4G网络、5G网络、蓝牙(Bluetooth)、Wi-Fi等无线或有线网络。其中，所述电子装置1通过所述网络3分别与一个或多个所述终端设备2通信连接。
参阅图2，是图1中电子装置1一可选的硬件架构的示意图，本实施例中，电子装置1可包括，但不仅限于，可通过系统总线相互通信连接的存储器11、处理器12、网络接口13。需要指出的是，图2仅示出了具有组件11-13的电子装置1，但是应理解的是，并不要求实施所有示出的组件，可以替代的实施更多或者更少的组件。
其中，所述存储器11至少包括一种类型的可读存储介质，所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如，SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘等。在一些实施例中，所述存储器11可以是所述电子装置1的内部存储单元，例如该电子装置1的硬盘或内存。在另一些实施例中，所述存储器11也可以是所述电子装置1的外部存储设备，例如该电子装置1上配备的插接式硬盘，智能存储卡(Smart Media Card,SMC)，安全数字(Secure Digital,SD)卡，闪存卡(Flash Card)等。当然，所述存储器11还可以既包括所述电子装置1的内部存储单元也包括其外部存储设备。本实施例中，所述存储器11通常用于存储安装于所述电子装置1的操作系统和各类应用软件，例如所述贷款风 险控制系统10的程序代码等。此外，所述存储器11还可以用于暂时地存储已经输出或者将要输出的各类数据。
所述处理器12在一些实施例中可以是中央处理器(Central Processing Unit，CPU)、控制器、微控制器、微处理器、或其他数据处理芯片。该处理器12通常用于控制所述电子装置1的总体操作，例如执行与所述终端设备2进行数据交互或者通信相关的控制和处理等。本实施例中，所述处理器12用于运行所述存储器11中存储的程序代码或者处理数据，例如运行所述贷款风险控制系统10等。
所述网络接口13可包括无线网络接口或有线网络接口，该网络接口13通常用于在所述电子装置1与其他电子设备之间建立通信连接。本实施例中，所述网络接口13主要用于通过所述网络3将所述电子装置1与一个或多个所述终端设备2相连，在所述电子装置1与一个或多个所述终端设备2之间建立数据传输通道和通信连接。
贷款风险控制系统10包括至少一个存储在所述存储器11中的计算机可读指令，该至少一个计算机可读指令可被所述处理器12执行，以实现本申请各实施例。
其中，上述贷款风险控制系统10被所述处理器12执行时实现如下步骤：
步骤S1，将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据。
本实施例中，首先获取一批种子用户，并对种子用户进行贷款风险等级的分级。例如，可由专家从用户数据库中挑选出一批具有代表性、普遍性的种子用户，并根据经验给种子用户进行贷款风险等级的分级，如可分成高风险、中风险、低风险等贷款风险等级。
获取预先建立的用户画像，并将挑选的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，以得到不同贷款风险等级的用户画像数据。其中，用户画像是建立在一系列真实数据之上的目标用户模型，是根据用户社会属性、生活习惯和消费行为等信息而抽象出的一个标签化的用户模型。构建用户画像的核心工作即是给用户贴“标签”，而标签是通过对用户信息分析而来的高度精炼的特征标识。本实施例中的预设用户画像可以是直接调用已建立好的与贷款人员相关的用户画像，也可以是通过各种数据源(如贷款网站数据库、QQ、微博、微信、雪球、东方财富社交软件等)来建立用户画像，建立的用户画像中包括各种属性标签，如用户的社会属性标签和金融属性标签，如年龄、家庭状况、收入、职业、消费习惯、是否有过贷款记录、是否办过信用卡等等。
将不同贷款风险等级的种子用户与对应的包含属性标签的用户画像进行关联，即将种子用户数据中的各项属性如年龄、性别、喜好、 收入等与用户画像中对应的各种属性标签进行匹配，将匹配度较高的种子用户数据与用户画像进行关联，将用户画像中的所有属性标签赋予与之相关联的种子用户数据，形成关联后的用户画像数据。这样，使得本来仅有基本属性的种子用户数据也具有了用户画像中的各种属性标签如各种社会属性标签和金融属性标签等，以便于后续建立模型及更加准确地对推广贷款业务的导流应用程序上用户进行贷款风险等级的分级。
步骤S2，基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进行分级，分为不同贷款风险等级。
本实施例中，获取到关联后的不同贷款风险等级的用户画像数据，可基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型。在一种可选的实施方式中，可采用一对多法(one-versus-rest，简称OVR SVMs)来训练得到分级模型，训练时依次把某个类别的样本即某一个贷款风险等级的用户画像数据归为一类,其他剩余的样本归为另一类，这样k个类别的样本就构造出了k个支持向量机SVM，分类时将未知样本分类为具有最大分类函数值的那类。在另一种可选的实施方式中，还可采用一对一法(one-versus-one,简称OVO SVMs或者pairwise)来训练得到分级模型，其做法是在任意两类样本即两种贷款风险等级的用户画像数据之间设计一个SVM，因此k个类别的样本就需要设计k(k-1)/2个SVM，当对一个未知样本进行分类时，最后得票最多的类别即为该未知样本的类别。
在一种具体实施方式中，在采用一对多法即多类划分法(one vs rest)对不同贷款风险等级的用户画像数据进行训练得到分级模型时，可假定贷款风险等级包括四个等级a、b、c、d，则所述分级模型的训练过程如下：
在抽取训练集的时候分别抽取贷款风险等级a所对应的用户画像数据向量作为正集，贷款风险等级b、c、d所对应的用户画像数据向量作为负集，以得到第一训练集；抽取贷款风险等级b所对应的用户画像数据向量作为正集，抽取贷款风险等级a、c、d所对应的用户画像数据向量作为负集，以得到第二训练集；抽取贷款风险等级c所对应的用户画像数据向量作为正集，贷款风险等级a、b、d所对应的用户画像数据向量作为负集，以得到第三训练集；抽取贷款风险等级d所对应的用户画像数据向量作为正集，贷款风险等级a、b、c所对应的用户画像数据向量作为负集，以得到第四训练集；对这四个训练集分别进行训练，得到四个支持向量机SVM，作为贷款风险等级的分级模型。
在利用所述分级模型对预设应用程序上用户的贷款风险等级进行分级时，提取出预设应用程序上用户的用户数据(如用户在应用程序上注册时填写的姓名、性别、年龄、职业等基本属性数据，或用户在使用应用程序过程中产生的该用户感兴趣的内容、消费习惯等等)，并将用户数据向量化，得到对应的用户数据向量。分别利用所述分级模型中的四个支持向量机SVM对用户数据向量进行测试，得到预设应用程序上用户的用户数据向量在每个支持向量机SVM上的分类函数值，将具有最大分类函数值的支持向量机SVM对应的用户画像数据向量的贷款风险等级作为该用户的贷款风险等级。即将第一训练集、第二训练集、第三训练集、第四训练集这四个训练集分别进行训练，然后得到四个支持向量机SVM的训练结果文件，在后续测试的时候，把对应的测试向量分别利用这四个训练结果文件进行测试，最后每个测试都有一个结果，最终的结果便是这四个值中最大的一个。
步骤S3，根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。
本实施例中，利用所述分级模型将预设应用程序上各个用户分为对应的不同贷款风险等级后，可对各个用户进行风险评分。例如，预先设定不同的贷款风险等级包括高、中等偏上、中等偏下、低四个级别，则在各个用户中，可设定高、中等偏上、中等偏下贷款风险等级的用户其风险评分低，则只向风险评分低的用户展现预设贷款入口。在进行风险评分时，还可在用户的贷款风险等级基础上，结合预先设定好的业务规则、用户类别、用户标签、用户的当前定位地址等因素综合对其进行贷款风险评分，只向评分达到预设要求的用户展现预设贷款入口，对评分达不到预设要求的高风险用户则不展现预设贷款入口，以阻止高风险用户通过预设贷款入口办理贷款业务。在展现预设贷款入口时，可根据设备号在评分达到预设要求的用户的终端设备显示该预设应用程序界面时在预设位置如底部、顶端等APP界面关键位置展示预设的贷款入口，以方便引导评分达到预设要求的用户进行贷款。
与现有技术相比，本实施例通过将不同贷款风险等级的种子用户与用户画像进行关联，得到不同贷款风险等级的用户画像数据，并基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，利用所述分级模型将预设应用程序上的用户分为不同贷款风险等级；根据用户对应的不同贷款风险等级进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。由于能对预设应用程序上的用户进行贷款风险评分，仅对评分符合要求的用户展现预设应用程序上 的贷款入口，提高了在导流的预设应用程序上推广贷款业务的贷款成功率，避免评分不符合要求的用户即劣质客户也能随意通过贷款入口办理贷款，降低了贷款公司的风险，提升贷款用户的转化率。
在一可选的实施例中，在上述图2的实施例的基础上，所述贷款风险控制系统10被所述处理器12执行实现所述步骤S3时，还包括：
获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序上用户的风险评分M的公式为：
M＝x1*M1+y1*M2
其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
本实施例中还可综合考虑用户在预设应用程序上的行为数据，并根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，例如，可根据不同用户在预设应用程序上的消费行为数据将聚类为经常消费、偶尔消费、基本不消费等类别，或习惯大额消费、习惯小额消费等类别。根据实际应用的需要预先设定不同贷款风险等级对应的评分基数M1及其对应的权重系数x1，以及不同行为类别对应的评分基数M2及其对应的权重系数y1，综合计算得到用户的风险评分M。例如，可预设不同贷款风险等级如高、中等偏上、中等偏下、低对应的评分基数分别为1、0.8、0.6、0.4，贷款风险等级对应的权重系数x1为0.4。预设不同行为类别如经常消费、偶尔消费、基本不消费类别的评分基数分别为0.5、0.7、0.9，行为类别对应的权重系数为0.6。若预设应用程序上一用户确定其为高贷款风险等级，且根据该用户在预设应用程序上的行为数据该用户确定其为偶尔消费类别，则可计算得到该用户的风险评分M＝0.4*1+0.6*0.7＝0.82。依次计算得到预设应用程序上各个用户的风险评分，评分越高说明贷款的风险越高，因此，仅对评分达到预设要求如低于某一预设阈值的用户展现预设贷款入口。
本实施例中，在计算用户的风险评分时，可综合考虑用户的贷款风险等级以及用户在预设应用程序上的行为数据，能更准确全面的评价用户的贷款风险，以更加精准地进行贷款风险控制。
在一可选的实施例中，所述贷款风险控制系统10被所述处理器12执行实现所述步骤S3时，还包括：
获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预 设标签，则计算预设应用程序上用户的风险评分M的公式为：
M＝x3*M3+y4*M4
其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
本实施例中还可综合考虑用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签。例如，可根据用户的定位数据中经常出现的定位地址为奢侈品商场，且该用户对应的用户画像数据中的收入属性标签为高收入人群，则可为该用户设置对应的预设标签为“高消费人群”；若定位地址为普通商场，且该用户对应的用户画像数据中的收入属性标签为中等收入人群，则可为该用户设置对应的预设标签为“中等消费人群”，等等。根据实际应用的需要预先设定不同贷款风险等级对应的评分基数M3及其对应的权重系数x3，以及不同标签对应的评分基数M4及其对应的权重系数y4，综合计算得到用户的风险评分M。例如，可预设不同贷款风险等级如高、中等偏上、中等偏下、低对应的评分基数分别为1、0.8、0.6、0.4，贷款风险等级对应的权重系数x1为0.3。预设不同标签如高消费人群、中等消费人群、低消费人群的评分基数分别为0.3、0.5、0.9，行为类别对应的权重系数为0.7。若预设应用程序上一用户确定其为高贷款风险等级，且根据该用户在预设应用程序上的定位数据确定该用户的标签为“中等消费人群”，则可计算得到该用户的风险评分M＝0.3*1+0.5*0.7＝0.65。依次计算得到预设应用程序上各个用户的风险评分，评分越高说明贷款的风险越高，因此，仅对评分达到预设要求如低于某一预设阈值的用户展现预设贷款入口。
本实施例中，在计算用户的风险评分时，可综合考虑用户的贷款风险等级以及用户在预设应用程序上的定位数据，能更准确全面的评价用户的贷款风险，以更加精准地进行贷款风险控制。
如图3所示，图3为本申请贷款风险控制方法一实施例的流程示意图，该贷款风险控制方法包括以下步骤：
步骤S10，将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据。
本实施例中，首先获取一批种子用户，并对种子用户进行贷款风险等级的分级。例如，可由专家从用户数据库中挑选出一批具有代表性、普遍性的种子用户，并根据经验给种子用户进行贷款风险等级的分级，如可分成高风险、中风险、低风险等贷款风险等级。
获取预先建立的用户画像，并将挑选的不同贷款风险等级的种子 用户与预先建立的用户画像进行关联，以得到不同贷款风险等级的用户画像数据。其中，用户画像是建立在一系列真实数据之上的目标用户模型，是根据用户社会属性、生活习惯和消费行为等信息而抽象出的一个标签化的用户模型。构建用户画像的核心工作即是给用户贴“标签”，而标签是通过对用户信息分析而来的高度精炼的特征标识。本实施例中的预设用户画像可以是直接调用已建立好的与贷款人员相关的用户画像，也可以是通过各种数据源(如贷款网站数据库、QQ、微博、微信、雪球、东方财富社交软件等)来建立用户画像，建立的用户画像中包括各种属性标签，如用户的社会属性标签和金融属性标签，如年龄、家庭状况、收入、职业、消费习惯、是否有过贷款记录、是否办过信用卡等等。
将不同贷款风险等级的种子用户与对应的包含属性标签的用户画像进行关联，即将种子用户数据中的各项属性如年龄、性别、喜好、收入等与用户画像中对应的各种属性标签进行匹配，将匹配度较高的种子用户数据与用户画像进行关联，将用户画像中的所有属性标签赋予与之相关联的种子用户数据，形成关联后的用户画像数据。这样，使得本来仅有基本属性的种子用户数据也具有了用户画像中的各种属性标签如各种社会属性标签和金融属性标签等，以便于后续建立模型及更加准确地对推广贷款业务的导流应用程序上用户进行贷款风险等级的分级。
步骤S20，基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进行分级，分为不同贷款风险等级。
本实施例中，获取到关联后的不同贷款风险等级的用户画像数据，可基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型。在一种可选的实施方式中，可采用一对多法(one-versus-rest，简称OVR SVMs)来训练得到分级模型，训练时依次把某个类别的样本即某一个贷款风险等级的用户画像数据归为一类,其他剩余的样本归为另一类，这样k个类别的样本就构造出了k个支持向量机SVM，分类时将未知样本分类为具有最大分类函数值的那类。在另一种可选的实施方式中，还可采用一对一法(one-versus-one,简称OVO SVMs或者pairwise)来训练得到分级模型，其做法是在任意两类样本即两种贷款风险等级的用户画像数据之间设计一个SVM，因此k个类别的样本就需要设计k(k-1)/2个SVM，当对一个未知样本进行分类时，最后得票最多的类别即为该未知样本的类别。
在一种具体实施方式中，在采用一对多法即多类划分法(one vs rest)对不同贷款风险等级的用户画像数据进行训练得到分级模型时， 可假定贷款风险等级包括四个等级a、b、c、d，则所述分级模型的训练过程如下：
在抽取训练集的时候分别抽取贷款风险等级a所对应的用户画像数据向量作为正集，贷款风险等级b、c、d所对应的用户画像数据向量作为负集，以得到第一训练集；抽取贷款风险等级b所对应的用户画像数据向量作为正集，抽取贷款风险等级a、c、d所对应的用户画像数据向量作为负集，以得到第二训练集；抽取贷款风险等级c所对应的用户画像数据向量作为正集，贷款风险等级a、b、d所对应的用户画像数据向量作为负集，以得到第三训练集；抽取贷款风险等级d所对应的用户画像数据向量作为正集，贷款风险等级a、b、c所对应的用户画像数据向量作为负集，以得到第四训练集；对这四个训练集分别进行训练，得到四个支持向量机SVM，作为贷款风险等级的分级模型。
在利用所述分级模型对预设应用程序上用户的贷款风险等级进行分级时，提取出预设应用程序上用户的用户数据(如用户在应用程序上注册时填写的姓名、性别、年龄、职业等基本属性数据，或用户在使用应用程序过程中产生的该用户感兴趣的内容、消费习惯等等)，并将用户数据向量化，得到对应的用户数据向量。分别利用所述分级模型中的四个支持向量机SVM对用户数据向量进行测试，得到预设应用程序上用户的用户数据向量在每个支持向量机SVM上的分类函数值，将具有最大分类函数值的支持向量机SVM对应的用户画像数据向量的贷款风险等级作为该用户的贷款风险等级。即将第一训练集、第二训练集、第三训练集、第四训练集这四个训练集分别进行训练，然后得到四个支持向量机SVM的训练结果文件，在后续测试的时候，把对应的测试向量分别利用这四个训练结果文件进行测试，最后每个测试都有一个结果，最终的结果便是这四个值中最大的一个。
步骤S30，根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。
本实施例中，利用所述分级模型将预设应用程序上各个用户分为对应的不同贷款风险等级后，可对各个用户进行风险评分。例如，预先设定不同的贷款风险等级包括高、中等偏上、中等偏下、低四个级别，则在各个用户中，可设定高、中等偏上、中等偏下贷款风险等级的用户其风险评分低，则只向风险评分低的用户展现预设贷款入口。在进行风险评分时，还可在用户的贷款风险等级基础上，结合预先设定好的业务规则、用户类别、用户标签、用户的当前定位地址等因素综合对其进行贷款风险评分，只向评分达到预设要求的用户展现预设贷款入口，对评分达不到预设要求的高风险用户则不展现预设贷款入 口，以阻止高风险用户通过预设贷款入口办理贷款业务。在展现预设贷款入口时，可根据设备号在评分达到预设要求的用户的终端设备显示该预设应用程序界面时在预设位置如底部、顶端等APP界面关键位置展示预设的贷款入口，以方便引导评分达到预设要求的用户进行贷款。
与现有技术相比，本实施例通过将不同贷款风险等级的种子用户与用户画像进行关联，得到不同贷款风险等级的用户画像数据，并基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，利用所述分级模型将预设应用程序上的用户分为不同贷款风险等级；根据用户对应的不同贷款风险等级进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。由于能对预设应用程序上的用户进行贷款风险评分，仅对评分符合要求的用户展现预设应用程序上的贷款入口，提高了在导流的预设应用程序上推广贷款业务的贷款成功率，避免评分不符合要求的用户即劣质客户也能随意通过贷款入口办理贷款，降低了贷款公司的风险，提升贷款用户的转化率。
在一可选的实施例中，在上述实施例的基础上，所述步骤S30还包括：
获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序上用户的风险评分M的公式为：
M＝x1*M1+y1*M2
其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
本实施例中还可综合考虑用户在预设应用程序上的行为数据，并根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，例如，可根据不同用户在预设应用程序上的消费行为数据将聚类为经常消费、偶尔消费、基本不消费等类别，或习惯大额消费、习惯小额消费等类别。根据实际应用的需要预先设定不同贷款风险等级对应的评分基数M1及其对应的权重系数x1，以及不同行为类别对应的评分基数M2及其对应的权重系数y1，综合计算得到用户的风险评分M。例如，可预设不同贷款风险等级如高、中等偏上、中等偏下、低对应的评分基数分别为1、0.8、0.6、0.4，贷款风险等级对应的权重系数x1为0.4。预设不同行为类别如经常消费、偶尔消费、基本不消费类别的评分基数分别为0.5、0.7、0.9，行为类别对应的权重系数为0.6。若预设应用程序上一用户确定其为高贷款风险等级，且根据 该用户在预设应用程序上的行为数据该用户确定其为偶尔消费类别，则可计算得到该用户的风险评分M＝0.4*1+0.6*0.7＝0.82。依次计算得到预设应用程序上各个用户的风险评分，评分越高说明贷款的风险越高，因此，仅对评分达到预设要求如低于某一预设阈值的用户展现预设贷款入口。
本实施例中，在计算用户的风险评分时，可综合考虑用户的贷款风险等级以及用户在预设应用程序上的行为数据，能更准确全面的评价用户的贷款风险，以更加精准地进行贷款风险控制。
在一可选的实施例中，所述步骤S3还包括：
获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签，则计算预设应用程序上用户的风险评分M的公式为：
M＝x3*M3+y4*M4
其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
本实施例中还可综合考虑用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签。例如，可根据用户的定位数据中经常出现的定位地址为奢侈品商场，且该用户对应的用户画像数据中的收入属性标签为高收入人群，则可为该用户设置对应的预设标签为“高消费人群”；若定位地址为普通商场，且该用户对应的用户画像数据中的收入属性标签为中等收入人群，则可为该用户设置对应的预设标签为“中等消费人群”，等等。根据实际应用的需要预先设定不同贷款风险等级对应的评分基数M3及其对应的权重系数x3，以及不同标签对应的评分基数M4及其对应的权重系数y4，综合计算得到用户的风险评分M。例如，可预设不同贷款风险等级如高、中等偏上、中等偏下、低对应的评分基数分别为1、0.8、0.6、0.4，贷款风险等级对应的权重系数x1为0.3。预设不同标签如高消费人群、中等消费人群、低消费人群的评分基数分别为0.3、0.5、0.9，行为类别对应的权重系数为0.7。若预设应用程序上一用户确定其为高贷款风险等级，且根据该用户在预设应用程序上的定位数据确定该用户的标签为“中等消费人群”，则可计算得到该用户的风险评分M＝0.3*1+0.5*0.7＝0.65。依次计算得到预设应用程序上各个用户的风险评分，评分越高说明贷款的风险越高，因此，仅对评分达到预设要求如低于某一预设阈值的用户展现预设贷款入口。
本实施例中，在计算用户的风险评分时，可综合考虑用户的贷款 风险等级以及用户在预设应用程序上的定位数据，能更准确全面的评价用户的贷款风险，以更加精准地进行贷款风险控制。
此外，本申请还提供一种计算机可读存储介质，所述计算机可读存储介质存储有贷款风险控制系统，所述贷款风险控制系统可被至少一个处理器执行，以使所述至少一个处理器执行如上述实施例中的贷款风险控制方法的步骤，该贷款风险控制方法的步骤S10、S20、S30等具体实施过程如上文所述，在此不再赘述。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件来实现，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质(如ROM/RAM、磁碟、光盘)中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，空调器，或者网络设备等)执行本申请各个实施例所述的方法。
以上参照附图说明了本申请的优选实施例，并非因此局限本申请的权利范围。上述本申请实施例序号仅仅为了描述，不代表实施例的优劣。另外，虽然在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤。
本领域技术人员不脱离本申请的范围和实质，可以有多种变型方案实现本申请，比如作为一个实施例的特征可用于另一实施例而得到又一实施例。凡在运用本申请的技术构思之内所作的任何修改、等同替换和改进，均应在本申请的权利范围之内。

Claims (20)

  1. 一种电子装置，其特征在于，所述电子装置包括存储器、处理器，所述存储器上存储有可在所述处理器上运行的贷款风险控制系统，所述贷款风险控制系统被所述处理器执行时实现如下步骤：
    A、将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据；
    B、基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进行分级，分为不同贷款风险等级；
    C、根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。
  2. 如权利要求1所述的电子装置，其特征在于，贷款风险等级包括四个等级a、b、c、d，所述分级模型的训练过程如下：
    利用多类划分法对不同贷款风险等级的用户画像数据进行训练，在抽取训练集的时候分别抽取贷款风险等级a所对应的用户画像数据向量作为正集，贷款风险等级b、c、d所对应的用户画像数据向量作为负集，以得到第一训练集；抽取贷款风险等级b所对应的用户画像数据向量作为正集，抽取贷款风险等级a、c、d所对应的用户画像数据向量作为负集，以得到第二训练集；抽取贷款风险等级c所对应的用户画像数据向量作为正集，贷款风险等级a、b、d所对应的用户画像数据向量作为负集，以得到第三训练集；抽取贷款风险等级d所对应的用户画像数据向量作为正集，贷款风险等级a、b、c所对应的用户画像数据向量作为负集，以得到第四训练集；对这四个训练集分别进行训练，得到四个支持向量机SVM，作为贷款风险等级的分级模型。
  3. 如权利要求2所述的电子装置，其特征在于，所述利用所述分级模型对预设应用程序上的用户进行分级包括：
    在利用所述分级模型对预设应用程序上用户的贷款风险等级进行分级时，将预设应用程序上用户的用户数据向量分别利用所述分级模型中的四个支持向量机SVM进行测试，得到预设应用程序上用户的用户数据向量在每个支持向量机SVM上的分类函数值，将具有最大分类函数值的支持向量机SVM对应的贷款风险等级作为该用户的贷款风险等级。
  4. 如权利要求1所述的电子装置，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x1*M1+y1*M2
    其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
  5. 如权利要求2所述的电子装置，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x1*M1+y1*M2
    其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
  6. 如权利要求3所述的电子装置，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x1*M1+y1*M2
    其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
  7. 如权利要求1所述的电子装置，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x3*M3+y4*M4
    其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
  8. 如权利要求2所述的电子装置，其特征在于，所述根据预设 应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x3*M3+y4*M4
    其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
  9. 如权利要求3所述的电子装置，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x3*M3+y4*M4
    其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
  10. 一种贷款风险控制方法，其特征在于，所述贷款风险控制方法包括：
    步骤一、将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据；
    步骤二、基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进行分级，分为不同贷款风险等级；
    步骤三、根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。
  11. 如权利要求10所述的贷款风险控制方法，其特征在于，贷款风险等级包括四个等级a、b、c、d，所述分级模型的训练过程如下：
    利用多类划分法对不同贷款风险等级的用户画像数据进行训练，在抽取训练集的时候分别抽取贷款风险等级a所对应的用户画像数据向量作为正集，贷款风险等级b、c、d所对应的用户画像数据向量作为负集，以得到第一训练集；抽取贷款风险等级b所对应的用户画像数据向量作为正集，抽取贷款风险等级a、c、d所对应的用户画像数据向量作为负集，以得到第二训练集；抽取贷款风险等级c所对应的用户画像数据向量作为正集，贷款风险等级a、b、d所对应的 用户画像数据向量作为负集，以得到第三训练集；抽取贷款风险等级d所对应的用户画像数据向量作为正集，贷款风险等级a、b、c所对应的用户画像数据向量作为负集，以得到第四训练集；对这四个训练集分别进行训练，得到四个支持向量机SVM，作为贷款风险等级的分级模型。
  12. 如权利要求11所述的贷款风险控制方法，其特征在于，所述利用所述分级模型对预设应用程序上的用户进行分级包括：
    在利用所述分级模型对预设应用程序上用户的贷款风险等级进行分级时，将预设应用程序上用户的用户数据向量分别利用所述分级模型中的四个支持向量机SVM进行测试，得到预设应用程序上用户的用户数据向量在每个支持向量机SVM上的分类函数值，将具有最大分类函数值的支持向量机SVM对应的贷款风险等级作为该用户的贷款风险等级。
  13. 如权利要求10所述的贷款风险控制方法，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x1*M1+y1*M2
    其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
  14. 如权利要求11所述的贷款风险控制方法，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x1*M1+y1*M2
    其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
  15. 如权利要求12所述的贷款风险控制方法，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的行为数据，根据所述行为数据将预设应用程序上的用户进行聚类为不同行为类别，则计算预设应用程序 上用户的风险评分M的公式为：
    M＝x1*M1+y1*M2
    其中，M1为预设的不同贷款风险等级对应的评分基数，x1为贷款风险等级对应的权重系数；M2为预设的不同行为类别对应的评分基数，y1为行为类别对应的权重系数。
  16. 如权利要求10所述的贷款风险控制方法，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x3*M3+y4*M4
    其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
  17. 如权利要求11所述的贷款风险控制方法，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x3*M3+y4*M4
    其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
  18. 如权利要求12所述的贷款风险控制方法，其特征在于，所述根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分包括：
    获取用户在预设应用程序上的定位数据，并根据所述定位数据及该用户的贷款风险等级对应的用户画像数据为该用户设置对应的预设标签，则计算预设应用程序上用户的风险评分M的公式为：
    M＝x3*M3+y4*M4
    其中，M3为预设的不同贷款风险等级对应的评分基数，x3为贷款风险等级对应的权重系数；M4为预设的不同标签对应的评分基数，y4为标签对应的权重系数。
  19. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质上存储有贷款风险控制系统，所述贷款风险控制系统被处理器执行时实现如下步骤：
    将预设的不同贷款风险等级的种子用户与预先建立的用户画像进行关联，得到不同贷款风险等级的用户画像数据；
    基于不同贷款风险等级的用户画像数据训练得到贷款风险等级的分级模型，并利用所述分级模型对预设应用程序上的用户进行分级，分为不同贷款风险等级；
    根据预设应用程序上用户对应的不同贷款风险等级，并按预设评分规则对预设应用程序上的用户进行风险评分，只向评分达到预设要求的用户展现预设贷款入口。
  20. 如权利要求19所述的计算机可读存储介质，其特征在于，贷款风险等级包括四个等级a、b、c、d，所述分级模型的训练过程如下：
    利用多类划分法对不同贷款风险等级的用户画像数据进行训练，在抽取训练集的时候分别抽取贷款风险等级a所对应的用户画像数据向量作为正集，贷款风险等级b、c、d所对应的用户画像数据向量作为负集，以得到第一训练集；抽取贷款风险等级b所对应的用户画像数据向量作为正集，抽取贷款风险等级a、c、d所对应的用户画像数据向量作为负集，以得到第二训练集；抽取贷款风险等级c所对应的用户画像数据向量作为正集，贷款风险等级a、b、d所对应的用户画像数据向量作为负集，以得到第三训练集；抽取贷款风险等级d所对应的用户画像数据向量作为正集，贷款风险等级a、b、c所对应的用户画像数据向量作为负集，以得到第四训练集；对这四个训练集分别进行训练，得到四个支持向量机SVM，作为贷款风险等级的分级模型。
PCT/CN2018/076139 2017-09-30 2018-02-10 贷款风险控制方法、电子装置及可读存储介质 WO2019061989A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710916484.8A CN107767259A (zh) 2017-09-30 2017-09-30 贷款风险控制方法、电子装置及可读存储介质
CN201710916484.8 2017-09-30

Publications (1)

Publication Number Publication Date
WO2019061989A1 true WO2019061989A1 (zh) 2019-04-04

Family

ID=61266987

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/076139 WO2019061989A1 (zh) 2017-09-30 2018-02-10 贷款风险控制方法、电子装置及可读存储介质

Country Status (2)

Country Link
CN (1) CN107767259A (zh)
WO (1) WO2019061989A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111062444A (zh) * 2019-12-21 2020-04-24 湖南大学 信用风险预测方法、系统、终端及存储介质
CN111260189A (zh) * 2020-01-08 2020-06-09 平安银行股份有限公司 一种风险控制方法、装置、计算机系统及可读存储介质
CN111383101A (zh) * 2020-03-25 2020-07-07 深圳前海微众银行股份有限公司 贷后风险监控方法、装置、设备及计算机可读存储介质
CN112330321A (zh) * 2020-11-20 2021-02-05 北京嘀嘀无限科技发展有限公司 数据处理方法、装置、电子设备和计算机可读存储介质
CN113689289A (zh) * 2021-08-26 2021-11-23 天元大数据信用管理有限公司 一种基于银行风险控制的方法及设备
CN116074118A (zh) * 2023-03-07 2023-05-05 北京安胜华信科技有限公司 一种api访问控制方法、系统、智能终端及存储介质
CN116934464A (zh) * 2023-07-26 2023-10-24 广东企企通科技有限公司 基于小微企业的贷后风险监控方法、装置、设备及介质

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108763051B (zh) * 2018-04-06 2023-07-25 平安证券股份有限公司 电子装置、交易软件运行风险预警方法及存储介质
CN110197315B (zh) * 2018-04-08 2024-03-22 腾讯科技（深圳）有限公司 风险评估方法、装置及其存储介质
CN109345374B (zh) * 2018-09-17 2023-04-18 平安科技（深圳）有限公司 风险控制方法、装置、计算机设备和存储介质
CN109584048A (zh) * 2018-11-30 2019-04-05 上海点融信息科技有限责任公司 基于人工智能对申请者进行风险评级的方法和装置
CN109859051A (zh) * 2019-01-15 2019-06-07 国网电子商务有限公司 一种融资方法及装置
CN110163481A (zh) * 2019-04-19 2019-08-23 深圳壹账通智能科技有限公司 电子装置、用户风控审核系统测试方法及存储介质
CN110135701A (zh) * 2019-04-23 2019-08-16 北京淇瑀信息科技有限公司 控制规则的自动生成方法、装置、电子设备及可读介质
CN110310012B (zh) * 2019-06-04 2023-07-28 平安科技（深圳）有限公司 数据分析方法、装置、设备及计算机可读存储介质
CN112395331B (zh) * 2020-11-17 2023-10-10 平安科技（深圳）有限公司 信用卡客户的用户画像方法、装置、设备及介质
CN113240509B (zh) * 2021-05-18 2022-04-22 重庆邮电大学 一种基于多源数据联邦学习的贷款风险评估方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130035912A1 (en) * 2011-08-04 2013-02-07 Margines Edward Y Systems and methods of processing personality information
CN106021376A (zh) * 2016-05-11 2016-10-12 上海点荣金融信息服务有限责任公司 用于处理用户信息的方法和设备
CN106503873A (zh) * 2016-11-30 2017-03-15 腾云天宇科技（北京）有限公司 一种预测用户守约概率的方法、装置和计算设备
CN107038256A (zh) * 2017-05-05 2017-08-11 平安科技（深圳）有限公司 基于数据源的业务定制装置、方法及计算机可读存储介质
CN107203518A (zh) * 2016-03-16 2017-09-26 阿里巴巴集团控股有限公司 在线系统个性化推荐的方法、系统以及装置、电子设备

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104463673A (zh) * 2014-12-22 2015-03-25 中国科学技术大学苏州研究院 一种基于支持向量机的p2p网络贷款风险评估模型
KR101597939B1 (ko) * 2015-07-02 2016-02-25 한국기업데이터 주식회사 거시경제지표를 활용한 산업별 신용위험도 예측 장치 및 방법
CN106157122A (zh) * 2016-07-24 2016-11-23 国家电网公司 一种互联网环境下电力用户信用评价方法
CN106557882A (zh) * 2016-11-29 2017-04-05 国网山东省电力公司电力科学研究院 基于多维度风险评价指标的电力用户筛选方法及系统
CN106897918A (zh) * 2017-02-24 2017-06-27 上海易贷网金融信息服务有限公司 一种混合式机器学习信用评分模型构建方法
CN107203939A (zh) * 2017-05-26 2017-09-26 阿里巴巴集团控股有限公司 确定用户风险等级的方法及装置、计算机设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20130035912A1 (en) * 2011-08-04 2013-02-07 Margines Edward Y Systems and methods of processing personality information
CN107203518A (zh) * 2016-03-16 2017-09-26 阿里巴巴集团控股有限公司 在线系统个性化推荐的方法、系统以及装置、电子设备
CN106021376A (zh) * 2016-05-11 2016-10-12 上海点荣金融信息服务有限责任公司 用于处理用户信息的方法和设备
CN106503873A (zh) * 2016-11-30 2017-03-15 腾云天宇科技（北京）有限公司 一种预测用户守约概率的方法、装置和计算设备
CN107038256A (zh) * 2017-05-05 2017-08-11 平安科技（深圳）有限公司 基于数据源的业务定制装置、方法及计算机可读存储介质

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111062444A (zh) * 2019-12-21 2020-04-24 湖南大学 信用风险预测方法、系统、终端及存储介质
CN111062444B (zh) * 2019-12-21 2023-12-08 湖南大学 信用风险预测方法、系统、终端及存储介质
CN111260189A (zh) * 2020-01-08 2020-06-09 平安银行股份有限公司 一种风险控制方法、装置、计算机系统及可读存储介质
CN111260189B (zh) * 2020-01-08 2023-06-06 平安银行股份有限公司 一种风险控制方法、装置、计算机系统及可读存储介质
CN111383101A (zh) * 2020-03-25 2020-07-07 深圳前海微众银行股份有限公司 贷后风险监控方法、装置、设备及计算机可读存储介质
CN111383101B (zh) * 2020-03-25 2024-03-15 深圳前海微众银行股份有限公司 贷后风险监控方法、装置、设备及计算机可读存储介质
CN112330321A (zh) * 2020-11-20 2021-02-05 北京嘀嘀无限科技发展有限公司 数据处理方法、装置、电子设备和计算机可读存储介质
CN113689289A (zh) * 2021-08-26 2021-11-23 天元大数据信用管理有限公司 一种基于银行风险控制的方法及设备
CN113689289B (zh) * 2021-08-26 2024-04-30 天元大数据信用管理有限公司 一种基于银行风险控制的方法及设备
CN116074118A (zh) * 2023-03-07 2023-05-05 北京安胜华信科技有限公司 一种api访问控制方法、系统、智能终端及存储介质
CN116074118B (zh) * 2023-03-07 2023-06-13 北京安胜华信科技有限公司 一种api访问控制方法、系统、智能终端及存储介质
CN116934464A (zh) * 2023-07-26 2023-10-24 广东企企通科技有限公司 基于小微企业的贷后风险监控方法、装置、设备及介质

Also Published As

Publication number Publication date
CN107767259A (zh) 2018-03-06

Similar Documents

Publication Publication Date Title
WO2019061989A1 (zh) 贷款风险控制方法、电子装置及可读存储介质
WO2019062021A1 (zh) 在应用程序中推送贷款广告的方法、电子装置及介质
US10936672B2 (en) Automatic document negotiation
WO2019218514A1 (zh) 网页目标信息的提取方法、装置及存储介质
WO2019184217A1 (zh) 热点事件分类方法、装置及存储介质
US20190197119A1 (en) Language-agnostic understanding
CN109492222B (zh) 基于概念树的意图识别方法、装置及计算机设备
US10114819B2 (en) Optimizing machine translations for user engagement
US20140335483A1 (en) Language proficiency detection in social applications
US9704171B2 (en) Methods and systems for quantifying and tracking software application quality
US9665551B2 (en) Leveraging annotation bias to improve annotations
US20180089542A1 (en) Training Image-Recognition Systems Based on Search Queries on Online Social Networks
WO2019041520A1 (zh) 基于社交数据的金融产品推荐方法、电子装置及介质
TW201843622A (zh) 基於資料源的業務客製裝置、方法及電腦可讀儲存介質
US10606910B2 (en) Ranking search results using machine learning based models
US11397952B2 (en) Semi-supervised, deep-learning approach for removing irrelevant sentences from text in a customer-support system
US20150248409A1 (en) Sorting and displaying documents according to sentiment level in an online community
US10496751B2 (en) Avoiding sentiment model overfitting in a machine language model
US10147020B1 (en) System and method for computational disambiguation and prediction of dynamic hierarchical data structures
WO2019061664A1 (zh) 电子装置、基于用户上网数据的产品推荐方法及存储介质
US10592782B2 (en) Image analysis enhanced related item decision
US9330075B2 (en) Method and apparatus for identifying garbage template article
WO2016169411A1 (zh) 一种信息处理方法及其设备
US20220121668A1 (en) Method for recommending document, electronic device and storage medium
JP6307822B2 (ja) プログラム、コンピュータおよび訓練データ作成支援方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18861168

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC , EPO FORM 1205A DATED 28.09.2020.

122 Ep: pct application non-entry in european phase

Ref document number: 18861168

Country of ref document: EP

Kind code of ref document: A1