WO2018214933A1 - 确定用户风险等级的方法及装置、计算机设备 - Google Patents

确定用户风险等级的方法及装置、计算机设备 Download PDF

Info

Publication number
WO2018214933A1
WO2018214933A1 PCT/CN2018/088192 CN2018088192W WO2018214933A1 WO 2018214933 A1 WO2018214933 A1 WO 2018214933A1 CN 2018088192 W CN2018088192 W CN 2018088192W WO 2018214933 A1 WO2018214933 A1 WO 2018214933A1
Authority
WO
WIPO (PCT)
Prior art keywords
user
risk
level
determining
index
Prior art date
Application number
PCT/CN2018/088192
Other languages
English (en)
French (fr)
Inventor
杨帆
付歆
Original Assignee
阿里巴巴集团控股有限公司
杨帆
付歆
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 阿里巴巴集团控股有限公司, 杨帆, 付歆 filed Critical 阿里巴巴集团控股有限公司
Publication of WO2018214933A1 publication Critical patent/WO2018214933A1/zh
Priority to US16/690,949 priority Critical patent/US20200090268A1/en

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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • 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
    • 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
    • G06Q40/06—Asset management; Financial planning or analysis

Definitions

  • the present application relates to the field of big data technologies, and in particular, to a method and apparatus for determining a user risk level, and a computer device.
  • the platform needs to evaluate the user's risk level and use the assessed risk level of each user to support the operation of the business. For example, in the Internet investment and financing scenario, the platform recommended financial products to users should meet the user's risk level.
  • the Internet platform generally uses questionnaires to allow users to fill in the content related to the risk level assessment to determine the user's risk level index.
  • the questionnaire survey method is inefficient, and does not guarantee the user's content and its actual situation. Consistency makes it impossible to accurately determine the level of risk for each user.
  • the present application provides a method and apparatus, and a computer device for determining a user risk level.
  • a method for determining a user risk level comprising:
  • first user data of the user and second user data the first user data reflecting at least one user attribute related to a risk tolerance capability of the user, wherein the second user data is in the business involving the risk of the user Behavioral data generated;
  • a method for determining a user risk level including:
  • an apparatus for determining a user risk level including:
  • the first obtaining unit acquires first user data and second user data of the user, where the first user data reflects at least one user attribute related to the risk tolerance capability of the user, where the second user data is the user Behavioral data generated in a business involving risk;
  • a first determining unit according to the first user data, determining a first index for characterizing the risk tolerance capability of the user
  • a second determining unit according to the second user data, determining a second index for characterizing a degree of risk preference of the user
  • the risk level determining unit determines the user risk level of the user according to the first index and the second index.
  • a computer apparatus comprising:
  • a memory for storing processor executable instructions
  • the processor is configured to:
  • first user data of the user and second user data the first user data reflecting at least one user attribute related to a risk tolerance capability of the user, wherein the second user data is in the business involving the risk of the user Behavioral data generated;
  • the foregoing process determines the first index and/or the second index according to the acquired user data by acquiring user data, and determines the user's risk according to the first index and/or the second index. Level, the resulting user risk level is highly accurate and efficient.
  • FIG. 1 is a flowchart of a method for determining a user risk level according to an exemplary embodiment
  • 3 is a process of determining a set variable that is related to a user's risk preference level, according to an exemplary embodiment
  • FIG. 4 is a system architecture, according to an exemplary embodiment
  • FIG. 5 is a diagram showing a hardware structure of an electronic device according to an exemplary embodiment.
  • the present application seeks to find a way to quickly and accurately measure a user's acceptance or preference for various types of risks that may be faced, which can be achieved through big data techniques.
  • the user's risk level in the investment and financial management can be evaluated through two main aspects: First, the user's subjective risk preference, that is, the user is psychologically correct. Investment risk, volatility, loss of investment, etc., whether it is preference or dislike, and the degree of preference or disgust; second, the objective risk tolerance of the user, that is, the measurement of investment risk, the possible loss of investment, etc. The magnitude of the impact of life, or the user's life goals.
  • the user's subjective risk preference different users have different preferences for risk.
  • high-risk, high-return wealth management products such as stocks, funds, etc.
  • low-risk, low-return wealth management products such as third-party current fund management products such as Yu'ebao.
  • the Internet platform needs to evaluate the subjective risk preference of users to recommend appropriate financial products to users according to the degree of risk preference of users, or to evaluate whether financial products sold to users are suitable for the user. Users, etc.
  • the user risk level can be obtained by filling out a questionnaire form, and the questions in the questionnaire include: family composition, income status, risk preference type, and the like.
  • the survey method has at least one or more of the following drawbacks:
  • the main factors include: the content that the user fills in the questionnaire often does not match the actual situation of the user, and there is the possibility of fraud on the supervisor; or, for some questions on the questionnaire, the user does not know how to answer, for example, how much percentage the user can bear The loss of this kind of problem users do not know how to answer; and so on.
  • the form of the questionnaire is too simple.
  • the data proves that the results of the questionnaire are very different from the behaviors that users actually show.
  • the accuracy of the results obtained in the form of the questionnaire needs to be improved.
  • the present application proposes a method for determining the risk level of the user more accurately and efficiently. The following describes the technical solution by various embodiments. .
  • FIG. 1 shows a flow of a method for determining a user risk level provided by an exemplary embodiment.
  • the method can be applied to computer equipment (such as a platform server providing a wealth management business, a cloud computing platform, etc.).
  • computer equipment such as a platform server providing a wealth management business, a cloud computing platform, etc.
  • the method includes the following steps 101-104, wherein:
  • step 101 the first user data of the user and the second user data are obtained, the first user data reflecting at least one user attribute related to the risk tolerance capability of the user, the second user data being the user Behavioral data generated in a business involving risk.
  • the first user data it may be user data generated by the user in the process of using various types of APPs.
  • the user attributes reflected by such first user data may include, but are not limited to, the user's age, gender, family composition, life stage, income status, personal assets, family assets, loans, and the like.
  • the attribute characteristics of the above various types of user attributes may be directly obtained by the application content being filled in by the user, or may be obtained indirectly through calculation of various types of user data.
  • the user's income can be calculated by the flow of the bank card; the user's assets can be estimated by the property owned by the name and other assets, and so on.
  • the service may be various types of services that provide services for users through the form of the Internet, such as self-service payment and other life service businesses, investment and wealth management and other financial services.
  • an application APP that provides the above services can be developed, allowing users to participate in these services through the APP, and multiple risk-related services can be provided on the same APP.
  • risk-related services usually involve risks, including the following situations: 1 Users may face risks after participating in the business, such as: users may cause financial losses after participating in investment and wealth management business.
  • the user can generate various types of user data in the process of operating the above-mentioned types of services involving risks through the APP.
  • the user data may be behavior data corresponding to the user's operation behavior.
  • the user's operation behavior includes but is not limited to: the user searches for a certain type of information on the APP, and the user is in the user.
  • the user's various operational behaviors can occur at various stages of the investment, such as before the investment behavior occurs, during the investment, and after the investment behavior ends.
  • the above behavior data may include, but is not limited to, the content viewed by the user, the time at which the user's viewing action occurs (starting time or ending time), and the duration of the viewing action.
  • the user data may also be data reflected by other events related to the business. For example, the data involved in the user's traffic driving incident (including the number of violations, the type of violation, etc.), the data involved in the user's physical examination event (including the time of the physical examination, the content of the medical examination, etc.).
  • the generated user data can be stored in a database to obtain relevant user data when it is necessary to determine the user's risk preference.
  • step 101 After the above step 101 is completed, the process proceeds to step 102 and step 103.
  • a first index for characterizing the risk tolerance of the user is determined based on the first user data.
  • step 102 can be specifically implemented by the following process:
  • Step 1021 Determine, according to the first user data, an attribute feature of the user under each user attribute of the plurality of user attributes.
  • Step 1022 Determine, according to the attribute feature, a first index for characterizing the risk tolerance of the user.
  • the attribute feature may be input to the first machine classification model, and the output of the first machine classification model is determined to be used to characterize the risk tolerance of the user.
  • the first index may be input to the first machine classification model, and the output of the first machine classification model is determined to be used to characterize the risk tolerance of the user.
  • one or more intervals may be determined for each user attribute, and one attribute feature is corresponding to each interval.
  • the user attribute is a personal asset, and a plurality of intervals are set according to the amount: 0 to 500,000 RMB, 50 to 2 million RMB, and 2 to 10 million RMB.
  • the attribute characteristic corresponding to 0-500,000 RMB can be defined as “1” (representing a group with low wealth level), and the attribute characteristic corresponding to 5 to 2 million RMB can be defined as “2” (representing a group with medium wealth level).
  • the attribute characteristic corresponding to 2 to 10 million RMB can be defined as “3” (representing a crowd with a high level of wealth).
  • the attribute characteristics under each user attribute can be separately determined according to the obtained first user data.
  • the first index may be a risk tolerance level of the user.
  • the user's risk tolerance level can be divided into five categories: low, medium, medium, medium, and high. Among them, users with low level of wealth and old age and high pressure of life can be classified as “low”; users with high wealth and young and small life pressure can be classified as “high”; The remaining three categories are users between "low” and "high”.
  • the first index may also be a value (between 0 and 1) that characterizes the user's risk tolerance, wherein the larger the value, the higher the risk tolerance of the user.
  • the above first machine classification model can be obtained by training through a machine learning algorithm.
  • the influence coefficient corresponding to each user attribute may also be determined by human experience, and the weighted summation is performed by using the determined respective influence coefficients to calculate the final first index.
  • a second index for characterizing the degree of risk preference of the user is determined based on the second user data.
  • step 103 can be implemented by the following process:
  • Step 1031 Determine, according to the second user data, a feature value of each of the plurality of setting variables in the setting variable, wherein the setting variable includes at least one determining a degree of risk preference affecting the user. Set the variable.
  • the second user data generated in the risk-related business not all data can reflect the user's risk preference, that is, not all data is related to the user's risk preference.
  • the second user data is actually related to the user's risk preference level, which is the target data that needs to be obtained when determining the user risk preference.
  • a user's physical examination event can reflect the user's attitude toward health risks. According to conventional understanding, this can reflect the user's attitude toward other types of risks, and some data corresponding to the physical examination event may be compiled with the user's risk. There is a degree of relevance.
  • one or more set variables that affect the user's risk preference can be set. Take the user's information search behavior as an example. If the content searched by the user in the APP mostly includes terms such as “stock” or “fund”, or the type of financial product searched is “stock class” or “fund class”, then To a certain extent, it can reflect that the user prefers high risk (that is, the user has a high degree of preference for investment risk). Conversely, if the user frequently searches for low-risk financial products, it can reflect that the user prefers low risk ( That is, users have low preference for investment risk).
  • the setting variable corresponding to the search behavior is: the type to which the search content belongs, and correspondingly, for each content type, a feature value corresponding to the content type may be determined in advance (ie, the variable is set). Assignment).
  • the content type is divided into a high risk type, a medium risk type, and a low risk type, and the eigenvalue corresponding to the high risk type is 1, the eigenvalue corresponding to the medium risk type is 0.5, and the eigenvalue corresponding to the low risk type is 0.
  • the user's information viewing behavior as an example, user A needs to view 100 other financial products before purchasing a certain financial product X.
  • the set variable is: the number of financial products that the user viewed before the investment behavior occurred. There are many types of variables to be set, and this article will not list them one by one.
  • a plurality of candidate setting variables may be defined in advance, and relevant technical means are used to verify whether the candidate setting variables are related to the user's preference for investment risk, and finally select A set variable that is related to the degree of risk preference of the user.
  • relevant technical means are used to verify whether the candidate setting variables are related to the user's preference for investment risk, and finally select A set variable that is related to the degree of risk preference of the user.
  • the plurality of setting variables may include setting variables that have no influence on the degree of risk preference of the user or have low influence (or low correlation), for example, such setting variables
  • the influence coefficient is set to 0 or close to zero.
  • the user data generated by the user's operation during the use of the APP is usually a statistical value.
  • a plurality of statistical value intervals may be set in advance for each set variable, and the statistical value intervals are used to determine the target users in each Set the feature value under the variable. Taking the number of high-risk financial products viewed by the user before the investment as an example, three statistical value intervals can be defined in advance: 1 to 10, 10 to 20, 20 to 50, and the eigenvalues corresponding to the three statistical value intervals are defined.
  • the characteristic value of the set variable is 0.1; when a user is investing When the number of high-risk financial products previously viewed is between 10 and 20, the characteristic value of the set variable is 0.2; when the number of high-risk financial products viewed by a user before investing is between 20 and 50, The characteristic value of this set variable is 0.3.
  • the eigenvalues of other types of set variables can be determined.
  • non-investment risks include, but are not limited to, the risks faced by users in the profession, the risks faced by users in terms of physical health, the risks faced by users in sports, the risks faced by users when driving, and others.
  • the risks faced in the financial scenario include, but are not limited to, the risks faced by users in the profession, the risks faced by users in terms of physical health, the risks faced by users in sports, the risks faced by users when driving, and others. The risks faced in the financial scenario.
  • the setting variables may include: selecting a self-employed or a highly stable industry such as a bank government, or a frequency at which the user changes jobs; when the user faces a risk in physical health, the setting variables may include the user experience. Frequency, stability, or the situation of the user purchasing health care products; when the user is engaged in sports, setting variables may include: whether the user likes to engage in high-risk sports, such as mountain climbing, skiing, and whether the user likes to engage in low-risk sports, such as fishing.
  • the setting variables may include: the speed at which the user drives, whether it is often overspeed or the number of violations, etc.; when the user's other financial scenarios, the setting variables may include: whether the user purchases sufficient insurance to prevent the future, the user Preference for choosing credit card payment, early consumption, or savings card consumption.
  • the above-mentioned various types of risk-related user data can also be obtained by providing a back-end database corresponding to the APP of the related service.
  • One or more set variables can be designed for other non-investment risks, and the relevant technical means are used to verify whether each set variable is a set variable that is related to the user's risk preference level.
  • Step 1032 Input the feature value of the user under each set variable into the second machine classification model, and determine the output of the second machine classification model as the second level for characterizing the risk preference of the user. index.
  • an influence coefficient may be determined for each set variable, and the process of calculating the risk preference index is roughly: first multiplying the characteristic value of each set variable by the corresponding influence of the set variable. The coefficients are then added to the respective products, and the sum obtained by the addition is determined as the user's risk preference index.
  • the machine classification model may be pre-trained, then in step 103, the feature value of the user under each set variable is input into the machine classification model, and the output of the machine classification model is determined.
  • the risk preference index for the user is a feature value under each of the plurality of set variables, and the output of the machine classification model is a possibility that the user is classified as a high risk preference type.
  • the risk preference index corresponding to the "low risk preference type user” is equal to or infinitely close to 0
  • the risk preference index corresponding to the "high risk preference type user” is equal to or infinitely close to 1.
  • the machine classification model can be trained by the following process:
  • Step 11 Filter out a plurality of sample users, the plurality of sample users including a plurality of sample users of a high risk preference type and a plurality of sample users of a low risk preference type.
  • sample users who belong to the high-risk preference type usually show an attitude of not paying attention or even preference for risk or loss in the investment.
  • samples that fall into the category of low-risk preferences are often extremely risk-averse in investing and try to avoid losses.
  • the two types of samples have significant differences in behavior.
  • step 11 can be specifically implemented by the following process:
  • a sample user who determines that the user data meets the high risk preference rule is a sample user of the high risk preference type, and the user whose user data meets the low risk preference rule is determined as the sample user of the low risk preference type.
  • the definition of a rule does not depend on whether the user has bought a high-risk product.
  • the definitions of the rules covered in this paper are taken from psychology, behavioral finance, and related theories under decision science. For example, by examining the psychological state and actual behavior of the user in the face of loss, the type of risk preference to which the user belongs is defined.
  • a “high-risk preference rule” defined may be “don't care after losing, continue to buy”, for example: the user continues to lose money in the proportion of ⁇ 20%, and / or the loss amount ⁇ 500RMB, continue Purchase a certain number of high-risk products; define a “low-risk preference rule” that can be “disliked after losing money”: users frequently check the profitability of personal assets when the account is profitable, and when the account generates a large loss, I dare not look at the profitability of personal assets.
  • the risk preference type to which the user belongs is defined by examining the psychological state and actual behavior of the user under the fluctuating market.
  • the defined “low risk preference rule” is “more sensitive when fluctuating”: when the stock market is stable, the user does not care about his assets, but whenever the market fluctuates greatly (for example, by 1%), the user Log in frequently and view your assets.
  • a plurality of different "high risk preference rules” and a plurality of different "low risk preference rules” can be separately defined, and the rules and existing user data can be utilized. Filter out sample users who meet various rules and type tags with “high risk preference” or “low risk preference”.
  • step 11 can be implemented by the following process:
  • the users in the experimental application whose behaviors meet the high risk preference rules are determined as sample users of the high risk preference type.
  • a user who blows a balloon more than a certain number of thresholds a may be defined as a high risk preferred user, and a user who is lower than another quantity threshold b (set value) may be defined as Low preference users, defining users between a and b as ambiguous users.
  • set value a certain number of thresholds a
  • set value a user who is lower than another quantity threshold b
  • Low preference users defining users between a and b as ambiguous users.
  • the experimental games used to obtain samples can also be of other types, which are not listed here.
  • Step 12 Acquire feature values of each of the plurality of sample users under each of the preset plurality of set variables.
  • the feature value may be determined according to user data of each sample user.
  • the set variables here are pre-designed variables that may be related to risk preferences.
  • Step 13 Train the machine classification model according to the feature value of each sample user in each of the plurality of sample users and the risk preference type corresponding to each sample user; wherein the machine The input to the classification model is a feature value under each of the plurality of set variables, the output of which is the likelihood that the user is classified as a high risk preference type.
  • the machine learning methods used in the training model may include, but are not limited to, linear regression, logistic regression, and the like.
  • the feature value of the target user under each set variable can be input into the machine classification model to output the risk preference index of the target user.
  • the risk appetite degree of the user may be divided into multiple levels according to requirements, such as: low, medium, and high, and the level of the user's risk preference level is determined according to the output risk preference index. For example, when the risk preference index is between 0 and 0.3, the risk preference level is “low”, the risk preference index is between 0.3 and 0.6, the risk preference level is “medium”, and the risk preference index is between 0.6 and 1. At the time, the level of risk preference is “high”.
  • a set variable on how to determine the degree of risk preference affecting the user may be verified by the sample user identified above in an embodiment of the present application. As shown in FIG. 3, the set variables that affect the degree of risk preference of the user can be determined by the following process:
  • Step 21 Filter out a plurality of sample users, the plurality of sample users including a plurality of sample users of a high risk preference type and a plurality of sample users of a low risk preference type.
  • Step 22 For any set variable to be verified, obtain a feature value of each sample user of the plurality of sample users under the set variable to be verified.
  • Step 23 Using the feature value of each sample user under the set variable to be verified and the risk preference type corresponding to each sample user, verify whether the set variable to be verified is a risk affecting user's risk preference level Constant variable.
  • step 23 can be specifically implemented by the following process:
  • Step 231 Determine, according to a feature value of each sample user under the set variable to be verified, a feature value rule of the sample user of the plurality of high risk preference types under the set variable to be verified, and the A eigenvalue law of a plurality of low risk preference type sample users under the set variable to be verified.
  • the eigenvalue law includes: an average value obtained by averaging a plurality of eigenvalues, or a distribution interval of a plurality of eigenvalues, and the like.
  • Step 232 If the difference between the feature value law corresponding to the high risk preference type and the feature value law corresponding to the low risk preference type meets the set condition, determine the set variable to be verified as affecting the user.
  • the user value samples of the "high risk preference type” and “low risk preference type” will have a large difference in the eigenvalue law of the set variable, and vice versa.
  • a certain set of variables does not affect the user's risk preference.
  • the user value of the "high risk preference type” and “low risk preference type” will have little or even the same difference in the eigenvalue law of the set variable.
  • a setting condition for measuring the difference can be set to determine whether the difference in the eigenvalue law of the user sample of the "high risk preference type” and the "low risk preference type” satisfies the setting condition. To finalize the set variables that meet the criteria.
  • the set variable to be verified is “the number of financial products viewed by the user before the investment behavior occurs”, it is assumed that the eigenvalues of the eight user samples of the pre-screened “high risk preference type” under the set variable respectively for:
  • the defined setting conditions are: the mean value x of each feature value of the user sample of the "high risk preference type” under the set variable and the feature value of the user sample of the "low risk preference type” under the set variable
  • the difference between the mean y is greater than four.
  • step 23 can also be specifically implemented by the following process:
  • the distribution of the feature values greater than the set threshold value in the plurality of sample users is separately calculated, and whether the set variable is a set variable affecting the degree of risk preference of the user is determined according to the distribution situation.
  • the threshold is set to 5
  • the distribution of the eigenvalues greater than 5 is: sample users of 2 "high risk preference types” and 8 "low risk preference types"
  • the distribution of the feature values corresponding to the set variables on the two types of user samples has obvious unevenness, indicating that the set variables have a greater impact on the user risk preference, and can be determined as A set variable that affects the user's risk preference.
  • one or more set variables that affect the user's risk preference may be designed based on human experience.
  • step 104 After the above steps 102 and 103 are completed, the process proceeds to step 104.
  • a user risk level of the user is determined according to the first index and the second index.
  • the first index and the second index may be scores for reflecting risk tolerance and risk preference, respectively, such as between 0 and 1. Among them, usually, the greater the score, the higher the risk tolerance or the higher the risk preference.
  • step 104 is specifically implemented by the following process:
  • Determining the risk tolerance level of the user according to the first index Determining the risk tolerance level of the user according to the first index; wherein the risk tolerance of the user may be divided into multiple levels from low to high, and each level may correspond to a value interval regarding the first index.
  • a level of risk preference of the user Determining, according to the second index, a level of risk preference of the user; wherein, similarly, the degree of risk preference of the user may be divided into multiple levels from low to high, and each level may correspond to one for the second index. Value range.
  • a user risk level of the user according to a predetermined level correspondence table, wherein the level correspondence table is used to describe a correspondence between the risk tolerance level, the risk preference level, and the user risk level relationship.
  • the risk tolerance level and the risk preference level need to be merged to obtain a user risk level that can finally reflect the user's risk level in investment.
  • the higher the user's risk tolerance level or the higher the risk preference level the higher the user's user risk level.
  • the above level correspondence table may be determined by the following process:
  • the number of levels of each of the above levels can be artificially set according to actual needs.
  • the number of levels corresponding to each of the above levels is determined by the computer according to a predefined rule. For example, determining the number of levels related to each level according to the number of platform users, it may be defined that when the number of platform users is greater than a certain number, the number of levels is increased; or, the number of levels corresponding to the user risk level is not less than the above risk tolerance level and risk preference. The number of levels corresponding to the degree level, and so on.
  • a risk tolerance level and a risk preference level corresponding to each user risk level are determined, and the level correspondence table is obtained.
  • the risk tolerance level and the risk preference level corresponding to each user risk level can be separately determined.
  • the risk tolerance level and the risk preference level corresponding to each user risk level can be artificially determined, and can also be determined by the computer according to a predefined rule, for example, the risk of each user is determined.
  • the number of times a level appears in the table, the number of times the intermediate level appears in the table can be greater than the number of times the high level or low level appears, and so on.
  • the corresponding weights may be set for the “risk preference level” and the “risk tolerance capability” respectively, and each of the weights is calculated according to the pre-divided risk preference level and the risk tolerance level, and combined with the above weights.
  • a score corresponding to the combination of the risk preference level and the risk tolerance level (the score reflects the level of the final user risk level), and finally, each of the above scores may be determined according to each of the calculated points.
  • the process of determining the level correspondence table is not limited herein. Of course, the correspondence of the levels may not exist in the form of a table.
  • risk tolerance is the main factor, supplemented by risk preference, that is, when the risk tolerance level is the same, the higher the risk preference level, the higher the user risk level; when the risk preference level is the same, the risk tolerance level is higher. High, the higher the user risk level.
  • the user risk level can be divided into 7 levels of 0-6. Among them, “0” represents the user with the lowest risk level, and the risk preference is the lowest and the risk tolerance is the lowest. “6” represents the user with the highest risk level, and has the highest risk tolerance and the highest level of risk preference.
  • the above process of calculating the user risk level may be performed every other specific period of time (such as daily), and the latest user data is obtained every day to determine the user risk level, and the data can be updated in time.
  • a method of determining a user risk level including:
  • User data of the user for reflecting at least one user attribute is obtained, the user attribute being related to the risk tolerance of the user.
  • a first index for characterizing the risk tolerance of the user is determined based on the attribute characteristics.
  • only the user data for determining the risk tolerance of the user may be acquired, and the risk tolerance of the user is determined based on the user data, and finally the user risk level is determined according to the first index.
  • the foregoing process determines the first index and/or the second index according to the acquired user data by acquiring user data, and determines the user's risk according to the first index and/or the second index. Level, the resulting user risk level is highly accurate and efficient. And, it can also ensure the timely update of data.
  • the system may include: a user equipment 100, a server 300 interacting with the user equipment, a first database 400 connected to the server 300, a device 200 for determining a user risk level, and a second database 500 and a third database 600.
  • the user equipment 100 can be installed with an APP having an investment and wealth management service
  • the server 300 is a platform server corresponding to the APP, and the platform server stores the second user data generated by the user in the business process involving the risk in the first database.
  • 400 is obtained by the device 200 for determining the user risk level.
  • the third database 600 can store first user data that can affect the risk tolerance of the user, and can be obtained by the device 200 for determining the user risk level, wherein the data in the third database 600 can be directly written by the server 300. It can also be collected and written by other application servers, and there is no limitation to this article.
  • the device 200 for determining the user risk level may be a virtual device existing on the server 300 in the form of program code. Of course, it should be noted that the device 200 may also be present on another computer device.
  • the device 200 acquires the required second user data from the first database 400, and extracts the feature values of the set variables, and inputs them to the pre-provided machine classification model, and outputs the first The second index (characterizing the user's risk appetite).
  • the device 200 can also obtain the required first user data from the third database 600, extract various attribute features, input to a preset machine classification model, and output a first index (characterizing the user's risk tolerance). .
  • the device 200 determines the user risk level according to the first index and the second index, and stores the user risk level in the second database 500, so as to prepare the user risk level for various application scenarios.
  • at least part of the databases in each of the above databases may also be the same database, and no limitation is imposed thereon.
  • FIG. 5 shows the structure of an electronic device provided by an exemplary embodiment.
  • the electronic device may be a computer device (such as a payment platform server or a financial platform server, etc.), and the electronic device may include a processor, an internal bus, a network interface, and a memory (including a memory and a non-volatile memory). ), of course, may also include the hardware needed for other businesses.
  • the processor reads the corresponding computer program from the non-volatile memory into memory and then runs.
  • the present application does not exclude other implementation manners, such as a logic device or a combination of software and hardware, etc., that is, the execution body of the following processing flow is not limited to each logical unit, and may be Hardware or logic device.
  • the foregoing apparatus 200 for determining a user risk level may include:
  • the obtaining unit 210 is configured to acquire first user data of the user and second user data, where the first user data reflects at least one user attribute related to a risk tolerance capability of the user, where the second user data is involved in the user Behavioral data generated in a risky business;
  • the first determining unit 220 determines, according to the first user data, a first index for characterizing the risk tolerance capability of the user;
  • the second determining unit 230 determines, according to the second user data, a second index for characterizing a degree of risk preference of the user;
  • the risk level determining unit 240 determines the user risk level of the user according to the first index and the second index.
  • the first determining unit 220 includes:
  • An attribute feature determining unit according to the first user data, determining an attribute feature of the user under each user attribute of the plurality of user attributes;
  • the first calculating unit inputs the attribute feature into the first machine classification model and determines an output of the first machine classification model as a first index for characterizing the risk tolerance of the user.
  • the second determining unit 230 includes:
  • the feature value determining unit determines, according to the second user data, a feature value of each of the plurality of set variables in the set variable, the set variable includes at least one determining a degree of risk preference affecting the user Setting variables;
  • a second calculating unit inputting a feature value of the user under each set variable into a second machine classification model, and determining an output of the second machine classification model as a degree of risk preference for characterizing the user Second index.
  • the risk level determining unit 240 includes:
  • a first level determining unit determining a risk tolerance level of the user according to the first index
  • a second level determining unit determining a risk preference level of the user according to the second index
  • a third level determining unit configured to determine a user risk level of the user according to a predetermined level correspondence table, wherein the level correspondence table is used to describe the risk tolerance level, the risk preference level, and the user Correspondence between risk levels.
  • the user risk level of the user is determined according to a correspondence between a predetermined risk tolerance level, a risk preference level, and a user risk level.
  • the method further includes:
  • the number determining unit determines the risk tolerance level, the risk preference level, and the user risk level level
  • the level correspondence table determining unit determines a risk tolerance level and a risk preference level corresponding to each user risk level based on the determined number of levels, and obtains the level correspondence table.
  • the risk-related business includes a service that has a risk of financial loss, and/or a service in which an associated event is at risk.
  • an apparatus for determining a user's risk tolerance capability including:
  • Obtaining a unit acquiring user data of the user for reflecting at least one user attribute, where the user attribute affects the risk tolerance capability of the user;
  • a third determining unit according to the user data, determining an attribute feature of the user under each user attribute of the plurality of user attributes
  • a fourth determining unit determines a first index for characterizing the risk tolerance of the user.
  • a computer storage medium having stored thereon a computer program, the computer program being executed by the processor to implement the following steps:
  • first user data of the user and second user data the first user data reflecting at least one user attribute
  • the second user data is behavior data generated by the user in a business involving risk
  • a computer storage medium having stored thereon a computer program, the computer program being executed by the processor to implement the following steps:
  • a first index for characterizing the risk tolerance of the user is determined based on the attribute characteristics.
  • a computer device including:
  • a memory for storing processor executable instructions
  • the processor is configured to:
  • first user data of the user and second user data the first user data reflecting at least one user attribute related to a risk tolerance capability of the user, wherein the second user data is in the business involving the risk of the user Behavioral data generated;
  • a computer device including:
  • a memory for storing processor executable instructions
  • the processor is configured to:
  • the system, device, module or unit illustrated in the above embodiments may be implemented by a computer chip or an entity, or by a product having a certain function.
  • a typical implementation device is a computer, and the specific form of the computer may be a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email transceiver, and a game control.
  • embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or a combination of software and hardware. Moreover, the invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
  • computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.
  • the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
  • the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
  • These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
  • the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.
  • a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
  • processors CPUs
  • input/output interfaces network interfaces
  • memory volatile and non-volatile memory
  • the memory may include non-persistent memory, random access memory (RAM), and/or non-volatile memory in a computer readable medium, such as read only memory (ROM) or flash memory.
  • RAM random access memory
  • ROM read only memory
  • Memory is an example of a computer readable medium.
  • Computer readable media includes both permanent and non-persistent, removable and non-removable media.
  • Information storage can be implemented by any method or technology.
  • the information can be computer readable instructions, data structures, modules of programs, or other data.
  • Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory. (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disk read only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, Magnetic tape cartridges, magnetic tape storage or other magnetic storage devices or any other non-transportable media can be used to store information that can be accessed by a computing device.
  • computer readable media does not include temporary storage of computer readable media, such as modulated data signals and carrier waves.
  • embodiments of the present application can be provided as a method, system, or computer program product.
  • the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment in combination of software and hardware.
  • the application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
  • the application can be described in the general context of computer-executable instructions executed by a computer, such as a program module.
  • program modules include routines, programs, objects, components, data structures, and the like that perform particular tasks or implement particular abstract data types.
  • the present application can also be practiced in distributed computing environments where tasks are performed by remote processing devices that are connected through a communication network.
  • program modules can be located in both local and remote computer storage media including storage devices.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Strategic Management (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • Theoretical Computer Science (AREA)
  • Economics (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Game Theory and Decision Science (AREA)
  • Technology Law (AREA)
  • Data Mining & Analysis (AREA)
  • Human Resources & Organizations (AREA)
  • Operations Research (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

一种确定确定用户风险等级的方法及装置、计算机设备，以提高用户风险等级的准确性。其中，确定用户风险等级的方法包括：获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。

Description

确定用户风险等级的方法及装置、计算机设备 技术领域
本申请涉及大数据技术领域，尤其涉及一种确定用户风险等级的方法及装置、计算机设备。
背景技术
随着互联网的发展，很多业务都可以通过互联网平台来实现。在一些业务的运营过程中，平台需要对用户的风险水平进行评估，并利用评估出的各个用户的风险水平来支撑业务的运营。例如，在互联网投资理财场景下，平台给用户推荐的理财产品应该符合用户的风险水平。
目前，互联网平台普遍采用问卷调查方式让用户填写与风险水平评估相关的内容，以确定出用户的风险水平指数，但是，问卷调查方式效率较低，且并不能保证用户填写的内容与其自身实际情况相符，导致无法准确地确定出每一用户的风险水平。
发明内容
有鉴于此，本申请提供一种确定用户风险等级的方法及装置、计算机设备。
为实现上述目的，本申请提供的技术方案如下：
根据本申请的第一方面，提出了一种确定用户风险等级的方法，包括：
获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二 指数；
根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
根据本申请的第二方面，提出了一种确定用户风险等级的方法，包括：
获取用户的用于反映至少一种用户属性的用户数据，所述用户属性与用户的风险承受能力相关；
根据所述用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数；
根据所述第一指数，确定所述用户的用户风险等级。
根据本申请的第三方面，提出了一种确定用户风险等级的装置，包括：
第一获取单元，获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
第一确定单元，根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
第二确定单元，根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
风险等级确定单元，根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
根据本申请的第四方面，提出了一种计算机设备，包括：
处理器；
用于存储处理器可执行指令的存储器；
所述处理器被配置为：
获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一 指数；
根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
通过以上技术方案可以看出，上述过程通过获取用户数据，并根据获取到的用户数据来确定第一指数和/或第二指数，并根据第一指数和/或第二指数来确定用户的风险等级，最终得到的用户风险等级准确性高，且效率高。
附图说明
图1为根据一示例性实施例示出的一种确定用户风险等级的方法的流程；
图2为根据一示例性实施例示出的一种训练机器分类模型的过程；
图3为根据一示例性实施例示出的一种确定与用户的风险偏好程度有相关性的设定变量的过程；
图4为根据一示例性实施例示出的一种系统架构；
图5为根据一示例性实施例示出的一种电子设备的硬件结构。
具体实施方式
本申请旨在寻找一种能快速、准确地衡量用户对可能面临的各类风险的接受程度或偏好程度的方法，该方法可以通过大数据技术来实现。以用户在投资理财过程中所面临的投资风险为例，可通过两个主要方面来评估用户在投资理财时的用户风险水平：其一，用户主观上对风险的偏好，即用户在心理上对投资风险、波动性、投资可能造成的损失等是否偏好或厌恶，以及偏好或厌恶的程度；其二，用户客观的风险承受能力，即衡量投资风险、投资可能造成的损失等因素对用户的实际生活、或用户的生活目标等产生的影响大小。其中，关于用户主观上对风险的偏好，不同的用户对风险的偏好不尽 相同，有的用户偏向于购买高风险、高回报的理财产品(如股票、基金等)，有的用户则偏向于购买低风险、低回报的理财产品(如余额宝等第三方活期资金理财产品)。为更好地服务于用户，互联网平台需要对用户主观上对风险的偏好程度进行评估，以根据用户的风险偏好程度，向用户推荐合适的金融产品，或评估销售给用户的金融产品是否适合该用户等。
在相关技术中，可以通过填写问卷的形式来获得用户风险水平，问卷中的问题包括：家庭组成、收入情况、风险偏好类型等。然而，问卷调查方式至少存在如下弊端中的一种或多种：
第一，无法获得与实际情况尽可能一致的结果。主要因素包括：用户在问卷上所填写的内容往往与用户自身实际情况不符，存在主管上造假的可能性；或者，对于问卷上的部分问题，用户不知如何回答，例如，询问用户能承受多少百分比的损失，这种问题用户并不知道如何回答；等等。
第二，问卷的形式过于简单，数据证明问卷调查的结果与用户真正表现出来的行为差异巨大。总之，问卷调查的形式获得的结果准确性有待提高，为提高准确性，本申请提出一种能够更为准确、高效地确定用户风险水平的方法，以下通过各种实施例来叙述这一技术方案。
图1示出了一示例性实施例提供的一种确定用户风险等级的方法的流程。该方法可应用于计算机设备(如提供投资理财业务的平台服务器、云计算平台等)。如图1所示，在一实施例中，该方法包括下述步骤101～104，其中：
在步骤101中，获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据。
关于第一用户数据，可以是用户在使用各类APP的过程中产生的用户数据。这类第一用户数据所反映的用户属性可以包括但不限于：用户的年龄，性别，家庭组成，所处的人生阶段，收入情况，个人资产，家庭资产，贷款情况等。上述各类用户属性的属性特征可以通过应用内容由用户填写的数据直接得到，也可以通过各类用户数据进行计算而间接得到。后者例如，用户 的收入，可以通过银行卡的流水情况来计算；用户的资产情况，可以通过名下所拥有的房产情况以及其他资产情况来估算，等等。
所述业务可为通过互联网形式来实现的各类为用户提供服务的业务，如：自助缴费等生活服务类业务、投资理财等金融类业务。一般地，可开发提供上述业务的应用APP，让用户通过APP来参与这些业务，并且，可在同一个APP上提供多种涉及风险的业务。其中，这类业务通常涉及到风险，包括如下情况：①用户参与业务后可能面临风险，如：用户参与投资理财业务后可能造成资金亏损。②与业务相关的特定事件存在风险，如：用户通过违章缴费业务进行自动缴费，与该业务相关的事件为交通驾驶事件，而交通驾驶事件是存在风险的；又比如，用户通过医疗服务业务来预约体检或预约看病等，体检事件或看病事件也涉及到用户在身体健康上所面临的风险；等等。
用户在通过APP针对上述涉及风险的各类业务进行操作的过程中，可以产生各类用户数据。在一实施例中，用户数据可以是与用户的操作行为对应的行为数据，以投资理财业务为例，用户的操作行为包括但不限于：用户在APP上针对某类信息的搜索行为，用户在APP上针对某类信息的查看行为，用户在APP上针对某类信息的评论行为，以及用户在APP上针对某类金融产品的购买行为。其中，用户的各种操作行为可以发生在投资的各个阶段，如：投资行为发生之前、投资中以及结束投资行为之后。上述行为数据可包括但不限于：用户查看的内容，用户的查看动作发生的时刻(起始时刻或终止时刻)，查看动作持续的时长等。在一实施例中，用户数据也可以是与业务相关的其他事件所反映的数据。如，用户的交通驾驶事件涉及的数据(包括违章次数，违章类型等)，用户的体检事件涉及的数据(包括体检的时间，体检的内容等)。产生的用户数据可被存储到数据库中，以便在需要确定用户的风险偏好时能获取到相关的用户数据。
在上述步骤101完成之后，进入步骤102以及步骤103。
在步骤102中，根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数。
用户的风险承受能力主要受用户所处的人生阶段以及用户的财富水平影响。在一实施例中，步骤102可具体通过如下过程来实现：
步骤1021：根据所述第一用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征。
步骤1022：根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数。
在一可选实施例中，在步骤1022中，可以将所述属性特征输入第一机器分类模型，并将所述第一机器分类模型的输出确定为用于表征所述用户的风险承受能力的第一指数。
其中，可以为每一种用户属性预先确定一个或多个区间，并为每个区间对应一个属性特征。例如，用户属性为个人资产，按照金额设定多个区间为：0～50万RMB，50～200万RMB，200～1000万RMB等。其中，可定义0～50万RMB对应的属性特征为“1”(代表财富水平低的人群)，可定义50～200万RMB对应的属性特征为“2”(代表财富水平中等的人群)，可定义200～1000万RMB对应的属性特征为“3”(代表财富水平高的人群)。以此类推，可以按照获取到的第一用户数据，分别确定各个用户属性下的属性特征。
在一实施例中，所述第一指数可为所述用户的风险承受能力等级。例如，可在风险承受能力这个维度，将用户的风险承受能力等级分为低、中低、中、中高、高五类。其中，财富水平低，且年纪大、生活压力大的用户可被分到“低”这一类；财富水平高，且年轻小、生活压力小的用户可被分到“高”这一类；其余三类为介于“低”和“高”之间的用户。当然，第一指数也可以是表征用户的风险承受能力的数值(可介于0～1之间)，其中该数值越大，表明用户的风险承受能力越高。
其中，上述第一机器分类模型可以通过机器学习算法来训练获得。
在其他实施例中，也可以通过人为经验来确定与每种用户属性对应的影响系数，并利用确定的各个影响系数进行加权求和，来计算得到最终的第一指数。
在步骤103中，根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数。
在一实施例中，步骤103可以通过如下过程来实现：
步骤1031：根据所述第二用户数据确定所述用户在多个设定变量中每个设定变量下的特征值，其中，所述设定变量中包括至少一个确定影响用户的风险偏好程度的设定变量。
实际上，涉及风险的业务中所产生的第二用户数据，并不是所有数据都能够反映用户的风险偏好程度，即并不是所有数据都与用户的风险偏好程度存在关联性。通常，只有部分第二用户数据是实际与用户的风险偏好程度存在关联性，这部分数据是在确定用户风险偏好时需要获取的目标数据。例如，用户的体检事件可以反映出用户在面临健康风险时的态度，按照常规理解，这可以反映出用户对其他类型风险的态度，则与体检事件对应的某些数据可能与用户的风险编好程度存在关联性。
为此，可以设定好一个或多个能够影响到用户的风险偏好程度的设定变量。以用户的信息搜索行为为例，如果用户在APP内搜索的内容大多包含“股票”或“基金”等词条，或搜索的金融产品的类型为“股票类”或“基金类”，则在一定程度上可以反映出该用户偏好于高风险(即用户对投资风险的偏好程度高)，反之，如果用户经常搜索的内容是低风险的金融产品，则可以反映出该用户偏好于低风险(即用户对投资风险的偏好程度低)。在该例子中，上述搜索行为对应的设定变量便为：搜索内容所属的类型，相应地，可以针对每一种内容类型，预先确定一个与该内容类型对应的特征值(即设定变量的赋值)。例如：将内容类型分为高风险类型、中风险类型及低风险类型，与高风险类型对应的特征值为1，与中风险类型对应的特征值为0.5，与低风险类型对应的特征值为0。以用户的信息查看行为为例，用户A在购买某一金融产品X之前，需要查看100个其他金融产品，用户B在购买某一金融产品X之前，需要查看10个其他金融产品，则表明用户A对投资风险是较为理性的，而用户B对投资风险则不太在意，也就是说，用户A对风险的偏好程 度要低于用户B对风险的偏好程度。在该例子中，设定变量为：用户在投资行为发生之前查看的金融产品的数目。设定变量的种类很多，本文不再一一作列举。
在一实施例中，可以预先定义出多种候选的设定变量，并通过相关技术手段来逐一验证这些候选的设定变量是否与用户对投资风险的偏好程度之间存在相关性，并最终选出与用户的风险偏好程度有相关性的设定变量。关于如何验证出与用户的风险偏好程度有相关性的设定变量的过程，将在下文予以详述。
需提及的是，所述多个设定变量中可以包括部分对用户的风险偏好程度没有影响或影响性较低(或相关性较低)的设定变量，例如，将这类设定变量的影响系数设定为0或接近于0。
用户在使用APP过程中的操作所产生的用户数据，通常是一种统计值。在一可选的实施例中，为了更加准确地计算出用户的风险偏好指数，可以预先为每一种设定变量设定多个统计值区间，并利用这些统计值区间来确定目标用户在各个设定变量下的特征值。以用户在投资之前查看的高风险类金融产品的数目为例，可以预先定义三个统计值区间：1～10，10～20，20～50，并定义这三个统计值区间对应的特征值分别为：0.1，0.2，0.3，则，当某用户在在投资之前查看的高风险类金融产品的数目介于1～10时，该设定变量的特征值为0.1；当某用户在在投资之前查看的高风险类金融产品的数目介于10～20时，该设定变量的特征值为0.2；当某用户在在投资之前查看的高风险类金融产品的数目介于20～50时，该设定变量的特征值为0.3。同理，按照这一规则可以确定出其他类型的设定变量的特征值。
能够想到的是，用户在生活中面临的风险种类(包括投资理财类风险及非投资类风险)很多，为了更加准确地确定出能够衡量用户的风险偏好程度的高低的风险偏好指数，需要尽可能获取用户在面临各类风险时的行为数据，并依据用户在面临各类风险时所作出的选择或操作，来确定用户的风险偏好程度的高低。举例来说，非投资类风险包括但不限于：用户在职业上面临的 风险、用户在身体健康状况上面临的风险、用户在从事体育运动所面临的风险、用户开车时所面临的风险、其他金融场景下所面临风险等。其中，用户面临职业风险时，设定变量可包括：选择自主创业还是银行政府等高稳定行业，或用户换工作的频率等；用户面临身体健康上的风险时，设定变量可包括用户体验的频率，稳定性，或用户购买保健用品的情况等；用户在从事体育运动时，设定变量可包括：用户是否喜欢从事高风险运动，例如登山，滑雪以及用户是否喜欢从事低风险运动，例如钓鱼；用户开车时面临的风险，设定变量可包括：用户开车的速度，是否经常超速或违章次数等；当用户的其他金融场景，设定变量可包括：用户是否购买充足的保险防范未来，用户偏好于选择信用卡支付，提前消费，还是储蓄卡消费等。上述各类风险相关的用户数据，也可以通过提供相关业务的APP对应的后台数据库来获取。
可针对其他非投资类风险设计出一个或多个设定变量，并通过相关技术手段来逐一验证每一个设定变量是否为与用户的风险偏好程度有相关性的设定变量。
步骤1032：将所述用户在每个设定变量下的特征值输入第二机器分类模型，并将所述第二机器分类模型的输出确定为用于表征所述用户的风险偏好程度的第二指数。
其中，在一实施例中，可为每个设定变量预先确定一个影响系数，则计算风险偏好指数的过程大致为：先将每个设定变量的特征值乘以该设定变量对应的影响系数，再将各个乘积相加，将相加所得的和值确定为用户的风险偏好指数。
在另一实施例中，可预先训练出机器分类模型，则在步骤103中，将所述用户在每个设定变量下的特征值输入机器分类模型，并将所述机器分类模型的输出确定为所述用户的风险偏好指数。上述机器分类模型的输入为所述多个设定变量中每个设定变量下的特征值，所述机器分类模型的输出为用户被分类为高风险偏好类型的可能性。其中，若将对风险的偏好程度最低的这一类用户定义为“低风险偏好类型的用户”，若将对风险的偏好程度最高的这 一类用户定义为“高风险偏好类型的用户”，则，“低风险偏好类型的用户”对应的风险偏好指数等于或无限接近于0，“高风险偏好类型的用户”对应的风险偏好指数等于或无限接近于1。其中，若某个用户的风险偏好指数越接近于0，则代表该用户属于“低风险偏好类型的用户”的可能性越高，若某个用户的风险偏好指数越接近于1，则代表该用户属于“高风险偏好类型的用户”的可能性越高。
图2为根据一示例性实施例示出的一种训练机器分类模型的过程。如图2所示，在一可选的实施例中，为提高准确性，可以通过以下过程训练出所述机器分类模型：
步骤11：筛选出多个样本用户，所述多个样本用户包括多个高风险偏好类型的样本用户和多个低风险偏好类型的样本用户。
其中，属于高风险偏好类型的样本用户通常是在投资中表现出对风险或损失不在乎甚至喜好的态度。相反，属于低风险偏好类型的样本用通常是在投资中极度厌恶风险，并极力避免损失的产生。一般地，两类样本在行为上具有明显的差异性。
关于如何筛选出多个样本用户的过程，又多种可行的实现方式，本文列举两种：
在一实施例中，步骤11可以具体通过如下过程来实现：
基于定义的高风险偏好规则和低风险偏好规则。将用户数据符合所述高风险偏好规则的用户确定为高风险偏好类型的样本用户，将用户数据符合所述低风险偏好规则的用户确定为低风险偏好类型的样本用户。与常规的定义不同，规则的定义不依赖于用户是否买了高风险产品。本文涉及的规则的定义取自于心理学、行为金融学以及决策科学下的相关理论。例如，通过考察用户在面临损失时的心理状态和实际行为来定义用户所归属的风险偏好类型。在该场景下，定义的一种“高风险偏好规则”可为“亏了后不在乎，继续买”，例如：用户在亏损的资金比例≥20％，和/或亏损金额≥500RMB时，仍继续购买一定数量的高风险产品；定义的一种“低风险偏好规则”可为“亏了以后就不 敢看”：用户在账户盈利时频繁地查看个人资产盈利情况，而账户产生大幅亏损时，就不敢再查看个人资产盈利情况。再比如，通过考察用户在波动行情下的心理状况和实际行为来定义用户所归属的风险偏好类型。在该场景下，定义的“低风险偏好规则”为“波动时更敏感”：在股票大盘平稳时，用户不关心自己的资产，但每当大盘大幅波动时(例如下跌1％)，用户就频繁地登录并查看自己的资产。当然，为提高筛选出的样本用户的准确性，可以分别定义出多种不同的“高风险偏好规则”以及多种不同的“低风险偏好规则”，并利用这些规则以及已有的用户数据，筛选出符合各类规则的样本用户，并打上“高风险偏好”或“低风险偏好”的类型标签。
在另一实施例中，步骤11可以具体通过如下过程来实现：
基于用于测试用户风险偏好的实验应用及定义的高风险偏好规则和低风险偏好规则，将在所述实验应用中的行为符合所述高风险偏好规则的用户确定为高风险偏好类型的样本用户，将在所述实验应用中的行为符合所述低风险偏好规则的用户确定为低风险偏好类型的样本用户。例如，开发一款“吹气球”的游戏，游戏中用户的任务是不断地吹气球，并获得与所吹气球大小正相关的金钱数额。与实际生活中的气球一样，如果用户将气球吹得过大(用户吹的次数越多，气球越大)，气球会爆炸，但是，气球被吹多大会爆是未知的。每一轮游戏用户都面临吹一次或离开的选择。如果用户选择吹气球，则将会有两种结果：①气球变大，获得的金钱更多，②气球吹爆了，已获得的金钱归零。而如果用户选择离开，则用户可以获得当前已经积累的金钱。在该游戏中，可以将吹气球的次数超过一定数量阈值a(设定的值)的用户定义成高风险偏好用户，而将低于另一数量阈值b(设定的值)的用户定义成低偏好偏好用户，将介于a、b之间的用户定义为不明确用户。当然，用于获得样本的实验游戏还可以为其它类型，本文不一一列举。
步骤12：获取所述多个样本用户中每个样本用户在预设的多个设定变量中每个设定变量下的特征值。其中，所述特征值可以根据各个样本用户的用户数据来确定。这里的设定变量是预先设计的各种可能与风险偏好相关的变 量。
步骤13：根据所述多个样本用户中每个样本用户在每个设定变量下的特征值，以及每个样本用户对应的风险偏好类型，训练出所述机器分类模型；其中，所述机器分类模型的输入为所述多个设定变量中每个设定变量下的特征值，所述机器分类模型的输出为用户被分类为高风险偏好类型的可能性。其中，训练模型所采用的机器学习方法可以包括但不限于：线性回归(linear regression)、逻辑回归(logistic regression)等。
在训练出备用的机器分类模型之后，便可以将所述目标用户在每个设定变量下的特征值输入机器分类模型，以输出所述目标用户的风险偏好指数。其中，可根据需要将用户的风险偏好程度划分为多个等级，如：低、中、高，并根据输出的风险偏好指数来确定用户的风险偏好程度的等级。例如，风险偏好指数介于0～0.3时，风险偏好程度的等级为“低”，风险偏好指数介于0.3～0.6时，风险偏好程度的等级为“中”，风险偏好指数介于0.6～1时，风险偏好程度的等级为“高”。
关于如何确定影响用户的风险偏好程度的设定变量，在本申请一实施例中，可以利用上述确定的样本用户来验证。如图3所示，可通过如下过程确定影响用户的风险偏好程度的设定变量：
步骤21：筛选出多个样本用户，所述多个样本用户包括多个高风险偏好类型的样本用户和多个低风险偏好类型的样本用户。
步骤22：对于任一待验证的设定变量，获取所述多个样本用户中每一样本用户在该待验证的设定变量下的特征值。
步骤23：利用每一样本用户在该待验证的设定变量下的特征值及每一样本用户对应的风险偏好类型，验证所述待验证的设定变量是否为影响用户的风险偏好程度的设定变量。
在一可选实施例中，该步骤23可以具体通过如下过程来实现：
步骤231：基于每一样本用户在该待验证的设定变量下的特征值，确定所述多个高风险偏好类型的样本用户在该待验证的设定变量下的特征值规律， 以及所述多个低风险偏好类型的样本用户在该待验证的设定变量下的特征值规律。例如：所述特征值规律包括：对多个特征值进行求平均运算所得的均值，或多个特征值的分布区间等。
步骤232：如果所述高风险偏好类型对应的特征值规律和所述低风险偏好类型对应的特征值规律之间的差异符合设定条件，将所述待验证的设定变量确定为影响用户的风险偏好程度的设定变量。
其中，对于影响用户的风险偏好程度的设定变量，“高风险偏好类型”和“低风险偏好类型”的用户样本在该设定变量上的特征值规律会呈现较大的差异，反之，若某设定变量对用户的风险偏好程度不产生影响，则“高风险偏好类型”和“低风险偏好类型”的用户样本在该设定变量上的特征值规律会差异较小甚至几乎相同。为此，可以设定用于衡量差异的设定条件，来判断“高风险偏好类型”和“低风险偏好类型”的用户样本在该设定变量上的特征值规律差异是否满足该设定条件，以最终确定出符合条件的设定变量。
例如，倘若待验证的设定变量为“用户在投资行为发生之前查看的金融产品的数目”，假设预先筛选的“高风险偏好类型”的8个用户样本在该设定变量下的特征值分别为：
{3、1、4、10、5、6、1、3}；
假设预先筛选的“低风险偏好类型”的8个用户样本在该设定变量下的特征值分别为：
{9、6、7、10、13、8、8、11}；
若定义的设定条件为：“高风险偏好类型”的用户样本在该设定变量下的各特征值的均值x和“低风险偏好类型”的用户样本在该设定变量下的各特征值的均值y之间的差值大于4。
通过计算，得出x＝4.15，y＝9，可见，满足上述设定条件，可确定“用户在投资行为发生之前查看的金融产品的数目”为影响用户的风险偏好程度的设定变量。
在另一可选的实施例中，上述步骤23也可以具体通过如下过程来实现：
分别统计大于设定阈值的特征值在所述多个样本用户中分布情况，并根据分布情况来确定该设定变量是否为影响用户的风险偏好程度的设定变量。
例如，在上述例子中，如设定阈值为5，则统计得出，大于5的特征值的分布情况为：2个“高风险偏好类型”的样本用户以及8个“低风险偏好类型”的样本用户，可见，该设定变量对应的特征值在两种类型的用户样本上的分布情况存在明显的不均匀，表明该设定变量对用户风险偏好产生较大的影响，可将其确定为影响用户的风险偏好程度的设定变量。
当然，在可选的其他实施例中，可以根据人为经验来设计一种或多种影响用户的风险偏好程度的设定变量。
在上述步骤102和步骤103完成之后，进入步骤104。
在步骤104中，根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
在一实施例中，上述第一指数和第二指数可以分别是用于反映风险承受能力以及风险偏好程度的分值(如介于0～1之间)。其中，通常，分值越大，可表示风险承受能力越高或风险偏好程度越高。
在一实施例中，则上述步骤104具体通过如下过程来实现：
根据所述第一指数确定所述用户的风险承受能力等级；其中，可以将用户的风险承受能力从低到高划分出多个等级，每个等级可以对应于一个关于第一指数的值区间。
根据所述第二指数确定所述用户的风险偏好程度等级；其中，同样地，可以将用户的风险偏好程度从低到高划分出多个等级，每个等级可以对应于一个关于第二指数的值区间。
根据预先确定的等级对应表，确定所述用户的用户风险等级，其中，所述等级对应表用以描述所述风险承受能力等级、所述风险偏好程度等级以及所述用户风险等级之间的对应关系。本申请实施例中，根据一般需求，需要将风险承受能力等级和风险偏好程度等级进行融合，以得到一个最终能够反映出用户在投资方面的风险水平的用户风险等级。通常，用户的风险承受能 力等级越高或风险偏好程度等级越高，该用户的用户风险等级也相应越高。
在一实施例中，可以通过如下过程来确定上述等级对应表：
分别确定风险承受能力等级、风险偏好程度等级以及用户风险等级的等级数。其中，可以根据实际需求，人为设定上述各个等级的等级数。或者，由计算机根据预定义的规则来确定上述各个等级对应的等级数。例如，根据平台用户数来确定各个等级相关的等级数，可以定义当平台用户数大于一定数量时，增加等级数；或者，定义用户风险等级对应的等级数不小于上述风险承受能力等级以及风险偏好程度等级对应的等级数，等等。
基于确定的等级数，确定与每一个用户风险等级相对应的风险承受能力等级和风险偏好程度等级，得到所述等级对应表。
在设定好上述各类等级相对应的等级数之后，可以分别确定出与每一个用户风险等级相对应的风险承受能力等级和风险偏好程度等级。其中，同样地，可以人为确定与每一个用户风险等级相对应的风险承受能力等级和风险偏好程度等级，也可以由计算机按照预定义规则来确定，其中，预定义规则例如：关于每个用户风险等级在表中出现的次数，中间等级在表中出现的次数可以大于高等级或低等级出现的次数，等等。
在其他实施例中，也可以分别为“风险偏好程度”和“风险承受能力”设定相应的权重，根据预先划分的风险偏好程度等级以及风险承受能力等级，并结合上述权重，计算出每一个风险偏好程度等级和风险承受能力等级的结合点所对应的分值(该分值反映出最终的用户风险等级的高低)，最终，可以根据计算出的各个分值，来确定出与上述每个结合点对应的用户风险等级。本文关于确定等级对应表的过程不作限制，当然，等级的对应关系可以不以表格的形式存在。
例如，上述等级对应表如下表1所示：
表1：
Figure PCTCN2018088192-appb-000001
其中，若以风险承受能力为主，以风险偏好程度为辅，即风险承受能力等级相同时，风险偏好程度等级越高，用户风险等级越高；风险偏好程度等级相同时，风险承受能力等级越高，用户风险等级越高。依据该原则，可将用户风险等级分为0～6这7个等级。其中，“0”代表用户风险等级最低的用户，其风险偏好程度最低，风险承受能力也最低。“6”代表用户风险等级最高的用户，其风险承受能力最高，风险偏好程度等级也最高。
在实际实现时，上述计算用户风险等级的过程可以是每隔一段特定时长(如每天)就执行一遍，每天都会获取最新的用户数据来确定用户风险等级，确保数据能够及时更新。
在另一实施例中，还提供一种确定用户风险等级的方法，包括：
获取用户的用于反映至少一种用户属性的用户数据，所述用户属性与用户的风险承受能力相关。
根据所述用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征。
根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数。
根据所述第一指数，确定所述用户的用户风险等级。
在本实施例中，可以仅获取用于确定用户的风险承受能力的用户数据，并依据这些用户数据确定出用户的风险承受能力，并最终根据第一指数来确 定出用户风险等级。
通过以上技术方案可以看出，上述过程通过获取用户数据，并根据获取到的用户数据来确定第一指数和/或第二指数，并根据第一指数和/或第二指数来确定用户的风险等级，最终得到的用户风险等级准确性高，且效率高。并且，也能保证数据的及时更新。
如图4所示，为一种系统架构。在一实施例中，该系统可以包括：用户设备100，与用户设备交互的服务器300，服务器300连接的第一数据库400，确定用户风险等级的装置200以及第二数据库500、第三数据库600。其中，用户设备100上可以安装具有投资理财业务的APP，服务器300为该APP对应的平台服务端，平台服务端将用户在参与涉及风险的业务过程中产生的第二用户数据存放于第一数据库400中，以备确定用户风险等级的装置200来获取。第三数据库600可以存放有能够影响用户的风险承受能力的第一用户数据，并且可以供确定用户风险等级的装置200来获取，其中第三数据库600中的数据可以是服务器300直接写入的，也可以是其他应用服务器来采集并写入的，对此本文均不作限制。其中，确定用户风险等级的装置200可以是存在于服务器300上的一种以程序代码形式存在的虚拟装置。当然，需说明的是，该装置200也可以存在于另外一个计算机装置上。当需要确定用户的风险等级时，该装置200从上述第一数据库400中获取到所需要的第二用户数据，并提取各设定变量的特征值，输入到预先提供的机器分类模型，输出第二指数(表征用户的风险偏好)。该装置200还可以从上述第三数据库600中获取到所需要的第一用户数据，并提取出各个属性特征，输入到预设的机器分类模型，输出第一指数(表征用户的风险承受能力)。最终，该装置200根据上述第一指数和第二指数确定用户风险等级并存放于第二数据库500中，以备各种应用场景调用用户风险等级。当然，上述各个数据库中的至少部分数据库也可以是同一个数据库，对此不作限制。
图5示出了一示例性实施例提供的一种电子设备的结构。如图5所示，所述电子设备可以为计算机设备(如支付平台服务器或理财平台服务器等)， 该电子设备可以包括处理器、内部总线、网络接口、存储器(包括内存以及非易失性存储器)，当然还可能包括其他业务所需要的硬件。处理器从非易失性存储器中读取对应的计算机程序到内存中然后运行。当然，除了软件实现方式之外，本申请并不排除其他实现方式，比如逻辑器件抑或软硬件结合的方式等等，也就是说以下处理流程的执行主体并不限定于各个逻辑单元，也可以是硬件或逻辑器件。
在一实施例中，上述确定用户风险等级的装置200可以包括：
获取单元210，获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
第一确定单元220，根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
第二确定单元230，根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
风险等级确定单元240，根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
在一可选的实施例中，所述第一确定单元220包括：
属性特征确定单元，根据所述第一用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
第一计算单元，将所述属性特征输入第一机器分类模型，并将所述第一机器分类模型的输出确定为用于表征所述用户的风险承受能力的第一指数。
在一可选的实施例中，所述第二确定单元230包括：
特征值确定单元，根据所述第二用户数据确定所述用户在多个设定变量中每个设定变量下的特征值，所述设定变量中包括至少一个确定影响用户的风险偏好程度的设定变量；
第二计算单元，将所述用户在每个设定变量下的特征值输入第二机器分类模型，并将所述第二机器分类模型的输出确定为用于表征所述用户的风险 偏好程度的第二指数。
在一可选的实施例中，所述风险等级确定单元240包括：
第一等级确定单元，根据所述第一指数确定所述用户的风险承受能力等级；
第二等级确定单元，根据所述第二指数确定所述用户的风险偏好程度等级；
第三等级确定单元，根据预先确定的等级对应表，确定所述用户的用户风险等级，其中，所述等级对应表用以描述所述风险承受能力等级、所述风险偏好程度等级以及所述用户风险等级之间的对应关系。根据预先确定的风险承受能力等级、风险偏好等级两者和用户风险等级之间的对应关系，确定所述用户的用户风险等级。
在一可选的实施例中，还包括：
等级数确定单元，分别确定风险承受能力等级、风险偏好程度等级以及用户风险等级的等级数；
等级对应表确定单元，基于确定的等级数，确定与每一个用户风险等级相对应的风险承受能力等级和风险偏好程度等级，得到所述等级对应表。
在一可选的实施例中，所述涉及风险的业务包括存在资金损失风险的业务、和/或相关联的事件存在风险的业务。
在一实施例中，还提供了一种确定用户风险承受能力的装置，包括：
获取单元，获取用户的用于反映至少一种用户属性的用户数据，所述用户属性影响所述用户的风险承受能力；
第三确定单元，根据所述用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
第四确定单元，根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数。
在一实施例中，还提供了一种计算机存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现如下步骤：
获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
在一实施例中，还提供了一种计算机存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现如下步骤：
获取用户的用于反映至少一种用户属性的用户数据，所述用户属性影响所述用户的风险承受能力；
根据所述用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数。
在一实施例中，还提供了一种计算机设备，包括：
处理器；
用于存储处理器可执行指令的存储器；
所述处理器被配置为：
获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
在一实施例中，还提供了一种计算机设备，包括：
处理器；
用于存储处理器可执行指令的存储器；
所述处理器被配置为：
获取用户的用于反映至少一种用户属性的用户数据，所述用户属性与用户的风险承受能力相关；
根据所述用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数；
根据所述第一指数，确定所述用户的用户风险等级。
本说明书中的各个实施例均采用递进的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于计算机设备实施例、或装置实施例、或计算机存储介质实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
上述实施例阐明的系统、装置、模块或单元，具体可以由计算机芯片或实体实现，或者由具有某种功能的产品来实现。一种典型的实现设备为计算机，计算机的具体形式可以是个人计算机、膝上型计算机、蜂窝电话、相机电话、智能电话、个人数字助理、媒体播放器、导航设备、电子邮件收发设备、游戏控制台、平板计算机、可穿戴设备或者这些设备中的任意几种设备的组合。
为了描述的方便，描述以上装置时以功能分为各种单元分别描述。当然，在实施本申请时可以把各单元的功能在同一个或多个软件和/或硬件中实现。
本领域内的技术人员应明白，本发明的实施例可提供为方法、系统、或计算机程序产品。因此，本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且，本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘 存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器，使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中，使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品，该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
在一个典型的配置中，计算设备包括一个或多个处理器(CPU)、输入/输出接口、网络接口和内存。
内存可能包括计算机可读介质中的非永久性存储器，随机存取存储器(RAM)和/或非易失性内存等形式，如只读存储器(ROM)或闪存(flash RAM)。内存是计算机可读介质的示例。
计算机可读介质包括永久性和非永久性、可移动和非可移动媒体可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括，但不限于相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编 程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带，磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质，可用于存储可以被计算设备访问的信息。按照本文中的界定，计算机可读介质不包括暂存电脑可读媒体(transitory media)，如调制的数据信号和载波。
还需要说明的是，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、商品或者设备不仅包括那些要素，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、商品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、商品或者设备中还存在另外的相同要素。
本领域技术人员应明白，本申请的实施例可提供为方法、系统或计算机程序产品。因此，本申请可采用完全硬件实施例、完全软件实施例或结合软件和硬件方面的实施例的形式。而且，本申请可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本申请可以在由计算机执行的计算机可执行指令的一般上下文中描述，例如程序模块。一般地，程序模块包括执行特定任务或实现特定抽象数据类型的例程、程序、对象、组件、数据结构等等。也可以在分布式计算环境中实践本申请，在这些分布式计算环境中，由通过通信网络而被连接的远程处理设备来执行任务。在分布式计算环境中，程序模块可以位于包括存储设备在内的本地和远程计算机存储介质中。
以上所述仅为本申请的实施例而已，并不用于限制本申请。对于本领域技术人员来说，本申请可以有各种更改和变化。凡在本申请的精神和原理之内所作的任何修改、等同替换、改进等，均应包含在本申请的权利要求范围之内。

Claims (15)

  1. 一种确定用户风险等级的方法，包括：
    获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
    根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
    根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
    根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
  2. 根据权利要求1所述的方法，所述根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数，包括：
    根据所述第一用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
    将所述属性特征输入第一机器分类模型，并将所述第一机器分类模型的输出确定为用于表征所述用户的风险承受能力的第一指数。
  3. 根据权利要求1所述的方法，所述根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数，包括：
    根据所述第二用户数据确定所述用户在多个设定变量中每个设定变量下的特征值，所述设定变量中包括至少一个确定影响用户的风险偏好程度的设定变量；
    将所述用户在每个设定变量下的特征值输入第二机器分类模型，并将所述第二机器分类模型的输出确定为用于表征所述用户的风险偏好程度的第二指数。
  4. 根据权利要求1所述的方法，所述根据所述第一指数和所述第二指数，确定所述用户的用户风险等级，包括：
    根据所述第一指数确定所述用户的风险承受能力等级；
    根据所述第二指数确定所述用户的风险偏好程度等级；
    根据预先确定的等级对应表，确定所述用户的用户风险等级，其中，所述等级对应表用以描述所述风险承受能力等级、所述风险偏好程度等级以及所述用户风险等级之间的对应关系。
  5. 根据权利要求4所述的方法，通过如下过程确定所述等级对应表：
    分别确定风险承受能力等级、风险偏好程度等级以及用户风险等级的等级数；
    基于确定的等级数，确定与每一个用户风险等级相对应的风险承受能力等级和风险偏好程度等级，得到所述等级对应表。
  6. 根据权利要求1所述的方法，所述涉及风险的业务包括存在资金损失风险的业务、和/或相关联的事件存在风险的业务。
  7. 一种确定用户风险等级的方法，包括：
    获取用户的用于反映至少一种用户属性的用户数据，所述用户属性与用户的风险承受能力相关；
    根据所述用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
    根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数；
    根据所述第一指数，确定所述用户的用户风险等级。
  8. 一种确定用户风险等级的装置，包括：
    获取单元，获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
    第一确定单元，根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
    第二确定单元，根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
    风险等级确定单元，根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
  9. 根据权利要求8所述的装置，所述第一确定单元包括：
    属性特征确定单元，根据所述第一用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
    第一计算单元，将所述属性特征输入第一机器分类模型，并将所述第一机器分类模型的输出确定为用于表征所述用户的风险承受能力的第一指数。
  10. 根据权利要求8所述的装置，所述第二确定单元包括：
    特征值确定单元，根据所述第二用户数据确定所述用户在多个设定变量中每个设定变量下的特征值，所述设定变量中包括至少一个确定影响用户的风险偏好程度的设定变量；
    第二计算单元，将所述用户在每个设定变量下的特征值输入第二机器分类模型，并将所述第二机器分类模型的输出确定为用于表征所述用户的风险偏好程度的第二指数。
  11. 根据权利要求8所述的装置，所述风险等级确定单元包括：
    第一等级确定单元，根据所述第一指数确定所述用户的风险承受能力等级；
    第二等级确定单元，根据所述第二指数确定所述用户的风险偏好程度等级；
    第三等级确定单元，根据预先确定的等级对应表，确定所述用户的用户风险等级，其中，所述等级对应表用以描述所述风险承受能力等级、所述风险偏好程度等级以及所述用户风险等级之间的对应关系。
  12. 根据权利要求11所述的装置，还包括：
    等级数确定单元，分别确定风险承受能力等级、风险偏好程度等级以及用户风险等级的等级数；
    等级对应表确定单元，基于确定的等级数，确定与每一个用户风险等级相对应的风险承受能力等级和风险偏好程度等级，得到所述等级对应表。
  13. 根据权利要求8所述的装置，所述涉及风险的业务包括存在资金损失风险的业务、和/或相关联的事件存在风险的业务。
  14. 一种计算机设备，包括：
    处理器；
    用于存储处理器可执行指令的存储器；
    所述处理器被配置为：
    获取用户的第一用户数据和第二用户数据，所述第一用户数据反映至少一种与用户的风险承受能力相关的用户属性，所述第二用户数据为所述用户在涉及风险的业务中产生的行为数据；
    根据所述第一用户数据，确定用于表征所述用户的风险承受能力的第一指数；
    根据所述第二用户数据，确定用于表征所述用户的风险偏好程度的第二指数；
    根据所述第一指数和所述第二指数，确定所述用户的用户风险等级。
  15. 一种计算机设备，包括：
    处理器；
    用于存储处理器可执行指令的存储器；
    所述处理器被配置为：
    获取用户的用于反映至少一种用户属性的用户数据，所述用户属性与用户的风险承受能力相关；
    根据所述用户数据，确定所述用户在多个用户属性中每个用户属性下的属性特征；
    根据所述属性特征，确定用于表征所述用户的风险承受能力的第一指数；
    根据所述第一指数，确定所述用户的用户风险等级。
PCT/CN2018/088192 2017-05-26 2018-05-24 确定用户风险等级的方法及装置、计算机设备 WO2018214933A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/690,949 US20200090268A1 (en) 2017-05-26 2019-11-21 Method and apparatus for determining level of risk of user, and computer device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710385586.1A CN107203939A (zh) 2017-05-26 2017-05-26 确定用户风险等级的方法及装置、计算机设备
CN201710385586.1 2017-05-26

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/690,949 Continuation US20200090268A1 (en) 2017-05-26 2019-11-21 Method and apparatus for determining level of risk of user, and computer device

Publications (1)

Publication Number Publication Date
WO2018214933A1 true WO2018214933A1 (zh) 2018-11-29

Family

ID=59905918

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/088192 WO2018214933A1 (zh) 2017-05-26 2018-05-24 确定用户风险等级的方法及装置、计算机设备

Country Status (4)

Country Link
US (1) US20200090268A1 (zh)
CN (1) CN107203939A (zh)
TW (1) TWI679604B (zh)
WO (1) WO2018214933A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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 (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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
CN107437198A (zh) 2017-05-26 2017-12-05 阿里巴巴集团控股有限公司 确定用户风险偏好的方法、信息推荐方法及装置
CN107767259A (zh) * 2017-09-30 2018-03-06 平安科技（深圳）有限公司 贷款风险控制方法、电子装置及可读存储介质
CN108280762A (zh) * 2018-01-19 2018-07-13 平安科技（深圳）有限公司 客户风险评级方法、服务器及计算机可读存储介质
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
CN109670962A (zh) * 2018-09-26 2019-04-23 深圳壹账通智能科技有限公司 基于大数据的理财产品推送方法、装置、设备及存储介质
CN109636081A (zh) * 2018-09-29 2019-04-16 阿里巴巴集团控股有限公司 一种用户安全意识检测方法和装置
CN110059920B (zh) * 2019-03-08 2021-08-06 创新先进技术有限公司 风险决策方法及装置
CN111523996A (zh) * 2020-04-21 2020-08-11 北京易点淘网络技术有限公司 一种审批方法及系统
CN113033938B (zh) * 2020-08-10 2024-04-26 深圳大学 确定资源分配策略的方法、装置、终端设备及存储介质
CN117455236A (zh) * 2020-09-11 2024-01-26 支付宝(杭州)信息技术有限公司 一种风测等级更新的方法、装置、设备及介质
US20220284419A1 (en) * 2021-03-05 2022-09-08 Dish Wireless L.L.C. Systems and methods for automatic asset transfer using smart contracts
CN113129153A (zh) * 2021-03-24 2021-07-16 银雁科技服务集团股份有限公司 风险评估方法及装置
CN113487326A (zh) * 2021-07-27 2021-10-08 中国银行股份有限公司 基于智能合约的交易限制参数设置方法及装置
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

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101685526A (zh) * 2008-09-28 2010-03-31 阿里巴巴集团控股有限公司 一种贷款准入评估方法和系统
CN102117469A (zh) * 2011-01-18 2011-07-06 中国工商银行股份有限公司 一种对信用风险进行评估的系统和方法
CN103208081A (zh) * 2013-03-01 2013-07-17 杭州智才广赢信息技术有限公司 一种投融资风险评估方法
CN105303445A (zh) * 2015-11-04 2016-02-03 中国农业大学 农业投融资平台风险评估装置及系统
CN105930983A (zh) * 2016-05-13 2016-09-07 中国建设银行股份有限公司 一种银行业资产管理系统
CN106127576A (zh) * 2016-07-01 2016-11-16 武汉泰迪智慧科技有限公司 一种基于用户行为的银行风险评估系统
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

Family Cites Families (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20030088489A1 (en) * 1999-12-13 2003-05-08 Optimizeusa.Com Automated investment advisory software and method
US20030065241A1 (en) * 2002-08-27 2003-04-03 Joerg Hohnloser Medical risk assessment system and method
WO2009059116A2 (en) * 2007-10-31 2009-05-07 Equifax, Inc. Methods and systems for providing risk ratings for use in person-to-person transactions
US20100030586A1 (en) * 2008-07-31 2010-02-04 Choicepoint Services, Inc Systems & methods of calculating and presenting automobile driving risks
US20110270780A1 (en) * 2010-03-24 2011-11-03 Gregory Bryn Davies Methods and systems for assessing financial personality
CN104318268B (zh) * 2014-11-11 2017-09-08 苏州晨川通信科技有限公司 一种基于局部距离度量学习的多交易账户识别方法
US10242019B1 (en) * 2014-12-19 2019-03-26 Experian Information Solutions, Inc. User behavior segmentation using latent topic detection
CN106503562A (zh) * 2015-09-06 2017-03-15 阿里巴巴集团控股有限公司 一种风险识别方法及装置
TWI584215B (zh) * 2015-12-31 2017-05-21 玉山商業銀行股份有限公司 監控可疑交易的方法
US9721296B1 (en) * 2016-03-24 2017-08-01 Www.Trustscience.Com Inc. Learning an entity's trust model and risk tolerance to calculate a risk score

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101685526A (zh) * 2008-09-28 2010-03-31 阿里巴巴集团控股有限公司 一种贷款准入评估方法和系统
CN102117469A (zh) * 2011-01-18 2011-07-06 中国工商银行股份有限公司 一种对信用风险进行评估的系统和方法
CN103208081A (zh) * 2013-03-01 2013-07-17 杭州智才广赢信息技术有限公司 一种投融资风险评估方法
CN105303445A (zh) * 2015-11-04 2016-02-03 中国农业大学 农业投融资平台风险评估装置及系统
CN105930983A (zh) * 2016-05-13 2016-09-07 中国建设银行股份有限公司 一种银行业资产管理系统
CN106127576A (zh) * 2016-07-01 2016-11-16 武汉泰迪智慧科技有限公司 一种基于用户行为的银行风险评估系统
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

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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
TWI679604B (zh) 2019-12-11
CN107203939A (zh) 2017-09-26
US20200090268A1 (en) 2020-03-19
TW201901578A (zh) 2019-01-01

Similar Documents

Publication Publication Date Title
WO2018214933A1 (zh) 确定用户风险等级的方法及装置、计算机设备
WO2018214935A1 (zh) 确定用户风险偏好的方法、信息推荐方法及装置
Fang et al. Cryptocurrency trading: a comprehensive survey
Li et al. Project success prediction in crowdfunding environments
Wu et al. Predicting winning price in real time bidding with censored data
US10185996B2 (en) Stock fluctuation prediction method and server
Chien Predicting patent litigation
US11250499B2 (en) Generating optimal strategy for providing offers
US20180046925A1 (en) Method, apparatus, and computer program product for determining closing metrics
TW202022769A (zh) 風險辨識模型訓練方法、裝置及伺服器
US20190171957A1 (en) System and method for user-level lifetime value prediction
US20190073699A1 (en) Matching visitors as leads to lead buyers
Burnie Exploring the interconnectedness of cryptocurrencies using correlation networks
CN106611375A (zh) 一种基于文本分析的信用风险评估方法及装置
US20170132706A1 (en) Preference Based Financial Tool System and Method
Collins et al. Mortgage modification and the decision to strategically default: A game theoretic approach
CN111242356A (zh) 一种财富走势预测方法、装置、设备及存储介质
CN115375177A (zh) 用户价值评估方法、装置、电子设备及存储介质
Rajwani et al. Measuring dependence between the USA and the Asian economies: A time-varying Copula approach
JP2019185595A (ja) 情報処理装置、情報処理方法、情報処理プログラム、判定装置、判定方法及び判定プログラム
Otabek et al. Twitter attribute classification with q-learning on bitcoin price prediction
Bitvai et al. Predicting peer-to-peer loan rates using Bayesian non-linear regression
CN109636432B (zh) 计算机执行的项目选择方法和装置
Aydemir et al. A dimension reduction approach to player rankings in European football
Fernandes et al. Decision-making simulator for buying and selling stock market shares based on twitter indicators and technical analysis

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: 18806093

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18806093

Country of ref document: EP

Kind code of ref document: A1