Loan risk control rating method and system based on big data

Abstract

The invention provides a loan risk control rating method and system based on big data, and the method comprises the steps: constructing a tag set corresponding to a user, and extracting static data information and dynamic data information related to the user from a multi-party data source; extracting identity data from the static data information, and obtaining an association score, a credit application score and an asset score from the static data information based on the identity data; extracting a feature data set from the dynamic data information to obtain a behavior weight and a behavior score, and further obtaining a consumption score and a browsing score; and generating a label weight, generating a loan score, and performing loan risk control rating based on the loan score. According to the method, corresponding content is added for the label set through the static data information and the dynamic data information, loan risk control rating is completed, the static data information and the dynamic data information are extracted based on multi-party data sources, diversified data sources are provided, timeliness, comprehensiveness and hierarchy of data are ensured, and accuracy of loan risk control rating is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data prediction, and in particular to a loan risk control rating method and system based on big data. Background Art

[0002] With the advancement of financial strategies, the credit economy has become a new economic growth point. Consumer credit, as the main manifestation of the credit economy, has developed rapidly and has gradually become a new source of profit growth and business transformation direction for financial institutions.

[0003] Loans refer to the money provided by financial institutions to consumers in advance for consumption, production and operation, etc., with the agreement to repay the principal and interest on time. With the popularization of consumer credit, its total amount and scale are increasing day by day, and loan defaults are also common. Therefore, it is more important for financial institutions to conduct risk control assessments on consumer loans.

[0004] Most current loan risk assessment methods rely on personal credit information, and the basis for evaluation mostly comes from the internal data of financial institutions. The data has shortcomings in timeliness, comprehensiveness and hierarchy, resulting in poor accuracy in loan risk assessment. Summary of the invention

[0005] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a loan risk control rating method and system based on big data, aiming to solve the technical problem that the prior art uses personal credit information formed based on the internal data of financial institutions to conduct loan risk control assessment, and its data has shortcomings in timeliness, comprehensiveness and hierarchy, resulting in poor accuracy of loan risk control assessment.

[0006] In order to achieve the above objectives, in a first aspect, an embodiment of the present application provides a loan risk control rating method based on big data, comprising the following steps: Constructing a tag set corresponding to the user, the tag set includes identity tags, stability tags, financial application tags, asset tags, commodity consumption tags, and media browsing tags, and extracting static data information and dynamic data information related to the user from multiple data sources, the dynamic data information includes a number of sub-data; Extracting identity data related to the identity tag from the static data information, wherein the identity data is uniquely associated with the user, and extracting a correlation score, a credit application score, and an asset score corresponding to the stability tag, the financial application tag, and the asset tag from the static data information based on the identity data; Performing a rationality check on the sub-data to determine whether the sub-data is erroneous data; if the sub-data is erroneous data, removing the sub-data to reduce the dynamic data information to the first adjustment data information; performing an integrity check on the sub-data in the first adjustment data information to obtain a missing rate of the sub-data in the first adjustment data information; determining whether the sub-data in the first adjustment data information needs to be removed based on the missing rate to reduce the first adjustment data information to the second adjustment data information; determining whether there are abnormal values ​​in the sub-data in the second adjustment data information; if there are abnormal values ​​in the sub-data in the second adjustment data information, removing the abnormal values; and performing data filling to obtain third adjustment data information; obtaining a keyword set corresponding to the commodity consumption tag and the media browsing tag; extracting a feature data set from the third adjustment data information based on the keyword set, wherein the feature data set includes a plurality of behavior data; Acquire a behavior weight and a behavior score corresponding to the behavior data, and generate a consumption score and a browsing score corresponding to the product consumption tag and the media browsing tag respectively based on the behavior weight and the behavior score; The tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag and the media browsing tag are obtained respectively, and a loan score is generated by combining the association score, the credit application score, the asset score, the consumption score and the browsing score, and a loan risk control rating is performed based on the loan score.

[0007] Compared with the prior art, the beneficial effects of the present invention are as follows: by distinguishing the static data information from the dynamic data information, the information extraction process can be accelerated based on the properties of different tags in the tag set, thereby improving the efficiency of loan risk control rating; after constructing the tag set, corresponding content is added to the tag set through the static data information and the dynamic data information to form a loan score corresponding to the user, thereby completing the loan risk control rating, and extracting the static data information and the dynamic data information from multiple data sources provides a diversified data source for loan risk control rating, thereby ensuring the timeliness, comprehensiveness and hierarchy of the data, thereby ensuring the accuracy of the loan risk control rating; by extracting the feature data set from the dynamic data information, after obtaining data from multiple data sources, a certain degree of data simplification is performed, thereby avoiding the reduction of data density caused by too much data, thereby further improving the efficiency of loan risk control rating.

[0008] Furthermore, the step of extracting the association score, the credit application score and the asset score corresponding to the stability tag, the financial application tag and the asset tag from the static data information based on the identity data includes: Acquire the number of terminals and the number of mailboxes corresponding to the identity data from the static data information, and determine the association score corresponding to the stability label based on the number of terminals and the number of mailboxes; Obtaining the number of applications and the number of application institutions corresponding to the identity data from the static data information, and determining a credit application score corresponding to the financial application tag based on the number of applications and the number of application institutions; The asset information corresponding to the identity data is obtained from the static data information, and the asset score corresponding to the asset tag is determined based on the asset information.

[0009] Furthermore, the step of judging whether to remove the sub-data in the first adjustment data information based on the missing rate is specifically: The missing rate of the sub-data in the first adjusted data information is compared with a missing rate threshold, and if the missing rate is greater than the missing rate threshold, the sub-data in the first adjusted data information is removed.

[0010] Furthermore, the step of determining whether there are abnormal values ​​in the sub-data in the second adjusted data information, and if the abnormal values ​​exist in the sub-data in the second adjusted data information, removing the abnormal values ​​and performing data filling includes: Splitting the sub-data in the second adjustment data information into a plurality of data pairs, wherein the data pairs include a time node and a data value corresponding to the time node, determining a value node based on the plurality of data values, and superimposing the value node with a threshold range to obtain a value interval; Determine whether the data value is within the value range, if the data value is not within the value range, determine that the data value is an abnormal value, and remove the abnormal value; The data change rate is obtained based on the time node and the data value, the data value corresponding to the time node adjacent to the time node corresponding to the abnormal value is selected as the reference value, and data is filled using the data change rate and the reference value.

[0011] Furthermore, the step of obtaining the behavior weight and behavior score corresponding to the behavior data includes: Segment the behavior data to obtain time segments, behavior types, link addresses, and behavior results; Generate a decay factor, a type weight and a link weight respectively based on the time segment, the behavior type and the link address; A behavior weight is generated by using the attenuation factor, type weight and link weight, and a behavior score is generated based on the behavior result.

[0012] Furthermore, the calculation formula of the behavior weight is: , in, represents the behavior weight, Attenuation factor, represents the type weight, Represents the link weight.

[0013] Furthermore, the identity data includes age data, and the step of obtaining the tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag respectively includes: Determine contribution factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, based on past loan scores, past association scores, past credit application scores, past asset scores, past consumption scores, and past browsing scores corresponding to past users; Determine, based on the age data, adjustment factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively; The tag weights respectively corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag are determined based on the contribution factor and the adjustment factor.

[0014] In a second aspect, an embodiment of the present application provides a loan risk control rating system based on big data, which is applied to the loan risk control rating method based on big data as described in the first aspect above, and the system includes: A construction module is used to construct a tag set corresponding to the user, the tag set includes identity tags, stability tags, financial application tags, asset tags, commodity consumption tags and media browsing tags, and extract static data information and dynamic data information related to the user from multiple data sources, the dynamic data information includes a number of sub-data; A first analysis module is configured to extract identity data associated with the identity tag from the static data information, wherein the identity data is uniquely associated with the user, and extract, based on the identity data, an association score, a credit application score, and an asset score corresponding to the stability tag, the financial application tag, and the asset tag from the static data information; a second analysis module, configured to perform a rationality test on the sub-data to determine whether the sub-data is erroneous data; if the sub-data is erroneous data, remove the sub-data to reduce the dynamic data information to the first adjustment data information; perform an integrity test on the sub-data in the first adjustment data information to obtain a missing rate of the sub-data in the first adjustment data information; determine whether the sub-data in the first adjustment data information needs to be removed based on the missing rate to reduce the first adjustment data information to the second adjustment data information; determine whether there are abnormal values ​​in the sub-data in the second adjustment data information; if there are abnormal values ​​in the sub-data in the second adjustment data information, remove the abnormal values; and perform data filling to obtain third adjustment data information; obtain a keyword set corresponding to the commodity consumption tag and the media browsing tag; and extract a feature data set from the third adjustment data information based on the keyword set, wherein the feature data set includes a plurality of behavior data; A third analysis module is used to obtain a behavior weight and a behavior score corresponding to the behavior data, and generate a consumption score and a browsing score corresponding to the product consumption tag and the media browsing tag respectively based on the behavior weight and the behavior score; An execution module is used to obtain the tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag and the media browsing tag, respectively, and generate a loan score in combination with the association score, the credit application score, the asset score, the consumption score and the browsing score, and perform loan risk control rating based on the loan score.

[0015] In a third aspect, an embodiment of the present application provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the loan risk rating method based on big data as described in the first aspect above is implemented.

[0016] In a fourth aspect, an embodiment of the present application provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the loan risk control rating method based on big data as described in the first aspect above is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a loan risk control rating method based on big data in the first embodiment of the present invention; Figure 2 This is a structural block diagram of a loan risk control rating system based on big data in the second embodiment of the present invention; The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0018] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0019] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0021] See also Figure 1 The first embodiment of the present invention provides a loan risk control rating method based on big data, comprising the following steps: S10: construct a tag set corresponding to the user, the tag set including identity tags, stability tags, financial application tags, asset tags, commodity consumption tags and media browsing tags, and extract static data information and dynamic data information related to the user from multiple data sources, the dynamic data information including several sub-data; The multi-party data sources include financial institution internal data sources, consumption platform data sources, loan platform data sources, transaction center data sources, terminal data sources and communication operator data sources. The static data information refers to data information with small data fluctuations within a fixed time period, which mainly includes identity information, gender, age, contact information, region, occupation, marital status, asset information, loan information, consumption level and other information with small fluctuations. The dynamic data information refers to data information with large data fluctuations, which mainly refers to the user's online operation behavior, such as browsing information on the homepage of the e-commerce website, browsing information on specific products on the website, searching information on products, purchasing information on the website and other constantly changing information. The sub-data refers to a specific behavior, such as: spending 200 yuan to buy a certain product on a certain platform.

[0022] S20: extracting identity data related to the identity tag from the static data information, wherein the identity data is uniquely associated with the user, and extracting a correlation score, a credit application score, and an asset score corresponding to the stability tag, the financial application tag, and the asset tag from the static data information based on the identity data; The identity data includes age and ID number.

[0023] The step S20 comprises: S210: Acquire the number of terminals and the number of mailboxes corresponding to the identity data from the static data information, and determine the association score corresponding to the stability label based on the number of terminals and the number of mailboxes; The more the number of terminals and mailboxes corresponding to the same user is, the lower the association score is.

[0024] S220: Obtaining the number of applications and the number of application institutions corresponding to the identity data from the static data information, and determining a credit application score corresponding to the financial application tag based on the number of applications and the number of application institutions.

[0025] The greater the number of applications and the number of application institutions corresponding to the same user, the lower the credit application score will be.

[0026] S230: Acquire asset information corresponding to the identity data from the static data information, and determine an asset score corresponding to the asset tag based on the asset information; The asset information includes real estate and car properties. The richer the asset information, the higher the asset score. The number of terminals, the number of mailboxes, the number of applications, the number of application institutions and the asset information are obtained and matched with the corresponding proportion division table to determine the association score, the credit application score and the asset score.

[0027] S30: Performing a rationality check on the sub-data to determine whether the sub-data is erroneous data; if the sub-data is erroneous data, removing the sub-data to reduce the dynamic data information to the first adjustment data information; performing an integrity check on the sub-data in the first adjustment data information to obtain a missing rate of the sub-data in the first adjustment data information; determining whether the sub-data in the first adjustment data information needs to be removed based on the missing rate to reduce the first adjustment data information to the second adjustment data information; determining whether there are abnormal values ​​in the sub-data in the second adjustment data information; if there are abnormal values ​​in the sub-data in the second adjustment data information, removing the abnormal values; and performing data filling to obtain third adjustment data information; obtaining a keyword set corresponding to the commodity consumption tag and the media browsing tag; extracting a feature data set from the third adjustment data information based on the keyword set; the feature data set includes a plurality of behavior data; It can be understood that if a certain sub-data is: spending -200 yuan to buy a certain product on a certain platform, it is extremely unreasonable, so the sub-data is determined to be erroneous data and is deleted.

[0028] The step of judging whether to remove the sub-data in the first adjustment data information based on the missing rate is specifically: The missing rate of the sub-data in the first adjusted data information is compared with a missing rate threshold, and if the missing rate is greater than the missing rate threshold, the sub-data in the first adjusted data information is removed.

[0029] The missing rate refers to the missing rate of important information in the sub-data. If a certain sub-data is a purchase record, but important information such as purchase amount, purchase time, purchase platform, etc. in the purchase record is missing, the sub-data needs to be eliminated.

[0030] The step of determining whether there is an abnormal value in the sub-data in the second adjusted data information, and if the abnormal value exists in the sub-data in the second adjusted data information, removing the abnormal value and performing data filling includes: Splitting the sub-data in the second adjustment data information into a plurality of data pairs, wherein the data pairs include a time node and a data value corresponding to the time node, determining a value node based on the plurality of data values, and superimposing the value node with a threshold range to obtain a value interval; Determine whether the data value is within the value range, if the data value is not within the value range, determine that the data value is an abnormal value, and remove the abnormal value; The data change rate is obtained based on the time node and the data value, the data value corresponding to the time node adjacent to the time node corresponding to the abnormal value is selected as the reference value, and data is filled using the data change rate and the reference value.

[0031] In this embodiment, the keyword set includes time, website, browsing, consumption, purchase, amount, etc. Based on the keyword set, corresponding result values ​​are extracted from the third adjustment data information and then combined into the behavior data. Several of the behavior data constitute the feature data set.

[0032] S40: Acquire a behavior weight and a behavior score corresponding to the behavior data, and generate a consumption score and a browsing score corresponding to the product consumption tag and the media browsing tag respectively based on the behavior weight and the behavior score; The step S40 comprises: S410: Segmenting the behavior data to obtain time segments, behavior types, link addresses, and behavior results; S420: Generate a decay factor, a type weight, and a link weight based on the time segment, the behavior type, and the link address; S430: Generate a behavior weight by using the attenuation factor, type weight and link weight, and generate a behavior score based on the behavior result; The time segment refers to the time when the behavior data occurred. The behavior weight of the behavior data gradually decreases with the passage of time, that is, the user's personal interests and behaviors will change at any time. Therefore, when the time is long, it is necessary to consider the problem of reduced compatibility between the behavior data and the user, so that the behavior score can be more accurately directed to the user. The behavior types include browsing type, comment type and purchase type, and the weights of different types are also different. The link address refers to the platform or website of purchase, and its weight will also be different depending on the popularity of different platforms or websites.

[0033] The calculation formula of the behavior weight is: , in, represents the behavior weight, Attenuation factor, represents the type weight, Represents the link weight.

[0034] S50: Obtaining tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, and generating a loan score in combination with the association score, the credit application score, the asset score, the consumption score, and the browsing score, and performing a loan risk control rating based on the loan score; The step S50 comprises: S510: Determine contribution factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, based on past loan scores, past association scores, past credit application scores, past asset scores, past consumption scores, and past browsing scores corresponding to past users; The acquisition method of the past association score, the past credit application score, the past asset score, the past consumption score and the past browsing score is consistent with the above steps and will not be repeated here. The past loan score can be obtained by rating based on the traditional loan risk control rating method according to the internal data source of the financial institution. With the past loan score as the result set, the past association score, the past credit application score, the past asset score, the past consumption score and the past browsing score as the evaluation set, the feature proportion analysis is performed to obtain the contribution factors corresponding to different tags.

[0035] S520: Determine adjustment factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, based on the age data; Understandably, for users of different ages, the weights corresponding to different tags are also different. Taking two users whose age data is 20 and 50 as an example, the weights of the commodity consumption tag and the media browsing tag corresponding to the 20-year-old user must be greater than the weights of the commodity consumption tag and the media browsing tag corresponding to the 50-year-old user.

[0036] S530: Determine the tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, based on the contribution factor and the adjustment factor.

[0037] The contribution factor is adjusted by the adjustment factor to form the label weight, and then the loan score is formed by superimposing different label weights with different scores.

[0038] By distinguishing the static data information from the dynamic data information, the information extraction process can be accelerated based on the properties of different tags in the tag set, thereby improving the efficiency of loan risk control rating. After constructing the tag set, corresponding content is added to the tag set through the static data information and the dynamic data information to form a loan score corresponding to the user, thereby completing the loan risk control rating. The static data information and the dynamic data information are extracted from multiple data sources, thereby providing a variety of data sources for loan risk control rating, ensuring the timeliness, comprehensiveness and hierarchy of the data, thereby ensuring the accuracy of the loan risk control rating. By extracting the feature data set from the dynamic data information, after obtaining data from multiple data sources, a certain degree of data simplification is performed to avoid a reduction in data density caused by too much data, thereby further improving the efficiency of loan risk control rating.

[0039] See also Figure 2 The second embodiment of the present invention provides a loan risk control rating system based on big data, which is applied to the loan risk control rating method based on big data in the above embodiment, and will not be repeated here. As used below, the terms "module", "unit", "subunit", etc. can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0040] The system comprises: A construction module 10 is used to construct a tag set corresponding to the user, the tag set including identity tags, stability tags, financial application tags, asset tags, commodity consumption tags and media browsing tags, and extract static data information and dynamic data information related to the user from multiple data sources, the dynamic data information including a number of sub-data; A first analysis module 20 is used to extract identity data related to the identity tag from the static data information, where the identity data is uniquely associated with the user, and extract an association score, a credit application score, and an asset score corresponding to the stability tag, the financial application tag, and the asset tag from the static data information based on the identity data; The first analysis module 20 includes: A first unit is configured to obtain the number of terminals and the number of mailboxes corresponding to the identity data from the static data information, and determine an association score corresponding to the stability label based on the number of terminals and the number of mailboxes; A second unit is used to obtain the number of applications and the number of application institutions corresponding to the identity data from the static data information, and determine a credit application score corresponding to the financial application tag based on the number of applications and the number of application institutions; A third unit is used to obtain asset information corresponding to the identity data from the static data information, and determine an asset score corresponding to the asset tag based on the asset information; The second analysis module 30 is used to perform a rationality test on the sub-data to determine whether the sub-data is erroneous data. If the sub-data is the erroneous data, the sub-data is removed to reduce the dynamic data information to the first adjustment data information, perform an integrity test on the sub-data in the first adjustment data information to obtain a missing rate of the sub-data in the first adjustment data information, determine whether the sub-data in the first adjustment data information needs to be removed based on the missing rate, to reduce the first adjustment data information to the second adjustment data information, determine whether there are abnormal values ​​in the sub-data in the second adjustment data information, if there are abnormal values ​​in the sub-data in the second adjustment data information, remove the abnormal values, and perform data filling to obtain third adjustment data information, obtain a keyword set corresponding to the commodity consumption tag and the media browsing tag, extract a feature data set from the third adjustment data information based on the keyword set, and the feature data set includes a plurality of behavior data; A third analysis module 40 is used to obtain a behavior weight and a behavior score corresponding to the behavior data, and generate a consumption score and a browsing score corresponding to the product consumption tag and the media browsing tag respectively based on the behavior weight and the behavior score; The third analysis module 40 includes: A fourth unit is used to obtain a keyword set corresponding to the commodity consumption tag and the media browsing tag, and extract a feature data set from the third adjustment data information based on the keyword set; The fifth unit is used to segment the behavior data to obtain time segments, behavior types, link addresses and behavior results; A sixth unit, for generating a decay factor, a type weight and a link weight respectively based on the time segment, the behavior type and the link address; A seventh unit is used to generate a behavior weight through the attenuation factor, the type weight and the link weight, and generate a behavior score based on the behavior result; An execution module 50 is used to obtain the tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag and the media browsing tag, respectively, and generate a loan score by combining the association score, the credit application score, the asset score, the consumption score and the browsing score, and perform loan risk control rating based on the loan score; The execution module 50 includes: An eighth unit is used to determine contribution factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, based on past loan scores, past association scores, past credit application scores, past asset scores, past consumption scores, and past browsing scores corresponding to past users; A ninth unit, for determining adjustment factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, based on the age data; The tenth unit is used to determine the tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag and the media browsing tag respectively based on the contribution factor and the adjustment factor.

[0041] The present invention also provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the loan risk control rating method based on big data as described in the above technical solution is implemented.

[0042] The present invention also provides a storage medium on which a computer program is stored. When the computer program is executed by a processor, the loan risk control rating method based on big data as described in the above technical solution is implemented.

[0043] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0044] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A loan risk control rating method based on big data, characterized in that: The following steps are involved: Constructing a tag set corresponding to the user, the tag set includes identity tags, stability tags, financial application tags, asset tags, commodity consumption tags, and media browsing tags, and extracting static data information and dynamic data information related to the user from multiple data sources, the dynamic data information includes a number of sub-data; Extracting identity data related to the identity tag from the static data information, wherein the identity data is uniquely associated with the user, and extracting a correlation score, a credit application score, and an asset score corresponding to the stability tag, the financial application tag, and the asset tag from the static data information based on the identity data; Performing a rationality check on the sub-data to determine whether the sub-data is erroneous data; if the sub-data is erroneous data, removing the sub-data to reduce the dynamic data information to the first adjustment data information; performing an integrity check on the sub-data in the first adjustment data information to obtain a missing rate of the sub-data in the first adjustment data information; determining whether the sub-data in the first adjustment data information needs to be removed based on the missing rate to reduce the first adjustment data information to the second adjustment data information; determining whether there are abnormal values ​​in the sub-data in the second adjustment data information; if there are abnormal values ​​in the sub-data in the second adjustment data information, removing the abnormal values; and performing data filling to obtain third adjustment data information; obtaining a keyword set corresponding to the commodity consumption tag and the media browsing tag; extracting a feature data set from the third adjustment data information based on the keyword set, wherein the feature data set includes a plurality of behavior data; Acquire a behavior weight and a behavior score corresponding to the behavior data, and generate a consumption score and a browsing score corresponding to the product consumption tag and the media browsing tag respectively based on the behavior weight and the behavior score; The tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag and the media browsing tag are obtained respectively, and a loan score is generated by combining the association score, the credit application score, the asset score, the consumption score and the browsing score, and a loan risk control rating is performed based on the loan score.

2. The big data-based loan risk control rating method according to claim 1, characterized in that: The step of extracting the association score, the credit application score and the asset score corresponding to the stability tag, the financial application tag and the asset tag from the static data information based on the identity data comprises: Acquire the number of terminals and the number of mailboxes corresponding to the identity data from the static data information, and determine the association score corresponding to the stability label based on the number of terminals and the number of mailboxes; Obtaining the number of applications and the number of application institutions corresponding to the identity data from the static data information, and determining a credit application score corresponding to the financial application tag based on the number of applications and the number of application institutions; The asset information corresponding to the identity data is obtained from the static data information, and the asset score corresponding to the asset tag is determined based on the asset information.

3. The big data-based loan risk control rating method according to claim 1, characterized in that: The step of judging whether to remove the sub-data in the first adjustment data information based on the missing rate is specifically: The missing rate of the sub-data in the first adjusted data information is compared with a missing rate threshold, and if the missing rate is greater than the missing rate threshold, the sub-data in the first adjusted data information is removed.

4. The big data-based loan risk control rating method according to claim 1, characterized in that: The step of determining whether there is an abnormal value in the sub-data in the second adjusted data information, and if the abnormal value exists in the sub-data in the second adjusted data information, removing the abnormal value and performing data filling includes: Splitting the sub-data in the second adjustment data information into a plurality of data pairs, wherein the data pairs include a time node and a data value corresponding to the time node, determining a value node based on the plurality of data values, and superimposing the value node with a threshold range to obtain a value interval; Determine whether the data value is within the value range, if the data value is not within the value range, determine that the data value is an abnormal value, and remove the abnormal value; The data change rate is obtained based on the time node and the data value, the data value corresponding to the time node adjacent to the time node corresponding to the abnormal value is selected as the reference value, and data is filled using the data change rate and the reference value.

5. The big data-based loan risk control rating method according to claim 1, characterized in that: The step of obtaining the behavior weight and behavior score corresponding to the behavior data comprises: Segment the behavior data to obtain time segments, behavior types, link addresses, and behavior results; Generate a decay factor, a type weight and a link weight respectively based on the time segment, the behavior type and the link address; A behavior weight is generated by using the attenuation factor, type weight and link weight, and a behavior score is generated based on the behavior result.

6. The big data-based loan risk control rating method according to claim 5, characterized in that: The calculation formula of the behavior weight is: ， in, represents the behavior weight, Attenuation factor, represents the type weight, Represents the link weight.

7. The big data-based loan risk control rating method according to claim 1, characterized in that: The identity data includes age data, and the step of obtaining the tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag respectively includes: Determine contribution factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively, based on past loan scores, past association scores, past credit application scores, past asset scores, past consumption scores, and past browsing scores corresponding to past users; Determine, based on the age data, adjustment factors corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag, respectively; The tag weights respectively corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag, and the media browsing tag are determined based on the contribution factor and the adjustment factor.

8. A loan risk control rating system based on big data, applied to the loan risk control rating method based on big data as claimed in any one of claims 1 to 7, characterized in that: The system comprises: A construction module is used to construct a tag set corresponding to the user, the tag set includes identity tags, stability tags, financial application tags, asset tags, commodity consumption tags and media browsing tags, and extract static data information and dynamic data information related to the user from multiple data sources, the dynamic data information includes a number of sub-data; A first analysis module is configured to extract identity data associated with the identity tag from the static data information, wherein the identity data is uniquely associated with the user, and extract, based on the identity data, an association score, a credit application score, and an asset score corresponding to the stability tag, the financial application tag, and the asset tag from the static data information; a second analysis module, configured to perform a rationality test on the sub-data to determine whether the sub-data is erroneous data; if the sub-data is erroneous data, remove the sub-data to reduce the dynamic data information to the first adjustment data information; perform an integrity test on the sub-data in the first adjustment data information to obtain a missing rate of the sub-data in the first adjustment data information; determine whether the sub-data in the first adjustment data information needs to be removed based on the missing rate to reduce the first adjustment data information to the second adjustment data information; determine whether there are abnormal values ​​in the sub-data in the second adjustment data information; if there are abnormal values ​​in the sub-data in the second adjustment data information, remove the abnormal values; and perform data filling to obtain third adjustment data information; obtain a keyword set corresponding to the commodity consumption tag and the media browsing tag; and extract a feature data set from the third adjustment data information based on the keyword set, wherein the feature data set includes a plurality of behavior data; A third analysis module is used to obtain a behavior weight and a behavior score corresponding to the behavior data, and generate a consumption score and a browsing score corresponding to the product consumption tag and the media browsing tag respectively based on the behavior weight and the behavior score; An execution module is used to obtain the tag weights corresponding to the stability tag, the financial application tag, the asset tag, the commodity consumption tag and the media browsing tag, respectively, and generate a loan score in combination with the association score, the credit application score, the asset score, the consumption score and the browsing score, and perform loan risk control rating based on the loan score.

9. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the big data-based loan risk control rating method according to any one of claims 1 to 7 is implemented.

Images