CN117291716A - Risk assessment method for goods money data - Google Patents
Risk assessment method for goods money data Download PDFInfo
- Publication number
- CN117291716A CN117291716A CN202311579699.7A CN202311579699A CN117291716A CN 117291716 A CN117291716 A CN 117291716A CN 202311579699 A CN202311579699 A CN 202311579699A CN 117291716 A CN117291716 A CN 117291716A
- Authority
- CN
- China
- Prior art keywords
- user
- evaluated
- clustering
- risk
- data
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 99
- 238000012502 risk assessment Methods 0.000 title claims abstract description 24
- 238000003064 k means clustering Methods 0.000 claims abstract description 61
- 230000000694 effects Effects 0.000 claims description 43
- 238000011156 evaluation Methods 0.000 claims description 42
- 230000008901 benefit Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Probability & Statistics with Applications (AREA)
- Development Economics (AREA)
- Technology Law (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the technical field of risk assessment of goods money data, in particular to a risk assessment method of goods money data. The method comprises the following steps: obtaining a first risk index based on the overdue time of each overdue of the money of the user to be evaluated, the total number of overdue times of the money, the total transaction number and the transaction number which are not paid up to the present; obtaining a repayment time change index based on the overdue duration of each overdue of the user to be evaluated and the duration between the overdue repayment time and the current time; clustering all users by adopting a K-means clustering algorithm and a DBSCAN clustering algorithm to obtain a preferred clustering result; obtaining a second risk index based on the preferred clustering result; and determining the risk level of the payment of the user to be evaluated based on the first risk index, the second risk index and the payment time change index. The invention improves the accuracy of the risk assessment of the goods money data.
Description
Technical Field
The invention relates to the technical field of risk assessment of goods money data, in particular to a risk assessment method of goods money data.
Background
Along with the rapid development of economy, the mode of co-win is widely used among enterprises, when different enterprises conduct transactions, the goods payment possibly has a certain risk, the goods payment risk possibly influences the normal operation of the enterprises, if the counterpart cannot timely repay in the transaction process, cash flow is likely to be interrupted, benefit loss is caused, the enterprises are likely to break when serious, therefore, before the transactions are conducted, the goods payment risk needs to be analyzed, the capability of the counterpart is estimated, whether the counterpart can timely repay or not is estimated, and corresponding measures are taken.
The existing risk assessment method is used for assessing the payment risk based on the recent repayment condition and the operation condition of a user, namely, the real-time behavior data of an individual user is matched with the historical data of the individual user to automatically analyze and evaluate, the method does not consider the relevance among enterprise transaction data, and the payment risk is not integrally controlled from different dimensions, so that the relevance among all aspects of the control is lacking, and the overall effect of the payment risk analysis and assessment is poor.
Disclosure of Invention
In order to solve the problem that the overall effect of the risk assessment of the goods money is poor due to incomplete automatic identification and control of risks in the aspect of individuals in the existing method, the invention aims to provide a risk assessment method of the goods money data, and the adopted technical scheme is as follows:
the invention provides a risk assessment method of goods money data, which comprises the following steps:
acquiring transaction data in the historical transaction process of all users and personal information of all users, wherein the transaction data comprises contracted repayment time, actual repayment time and goods money, and the personal information comprises registered funds and operation duration;
obtaining a first risk index of to-be-transacted data based on the overdue duration of each overdue of the to-be-evaluated user, the total number of overdue of the money, the total transaction number and the number of transactions which are not paid up to the present; obtaining a repayment time change index corresponding to the user to be evaluated based on the overdue duration of each overdue of the user to be evaluated and the duration between the overdue repayment time and the current time;
Clustering all users by adopting a K-means clustering algorithm and a DBSCAN clustering algorithm respectively, and obtaining a preferable clustering result of the K-means clustering based on the number of users in a clustering cluster where the users to be evaluated are located in the DBSCAN clustering result, the difference degree of transaction data of each user and the data to be transacted, the intra-class difference and the inter-class difference in the K-means clustering result; the K-means clustering algorithm is based on registered funds, operation duration and average amount of money in the historical transaction process of each user, and the DBSCAN clustering algorithm is based on transaction data of the users in the historical transaction process;
obtaining a second risk index of the data to be traded based on a first risk index of the transaction data of each user in the cluster where the user to be evaluated is located in the preferred cluster result and the difference degree of the transaction data of each user and the data to be traded; and determining the risk level of the payment of the user to be evaluated based on the first risk index, the second risk index and the repayment time change index of the data to be transacted.
Preferably, the obtaining a preferred clustering result of the K-means cluster based on the number of users in the cluster where the users to be evaluated are located in the DBSCAN clustering result, the degree of difference between the transaction data of each user and the data to be transacted, the intra-class difference and the inter-class difference in the K-means clustering result includes:
Marking the users in the cluster where the users to be evaluated are located in the K-means cluster result as target users, and obtaining the difference index of the transaction data in the cluster where the users to be evaluated are located in the DBSCAN cluster result based on the difference degree of the transaction data and the transaction data of each target user in the cluster where the transaction data of the users to be evaluated are located in the DBSCAN cluster result;
based on intra-class differences and inter-class differences in the K-means clustering result and difference indexes of transaction data in a clustering cluster where a user to be evaluated is located in the DBSCAN clustering result, obtaining a clustering effect evaluation index;
if the clustering effect evaluation index is larger than the clustering effect evaluation index threshold, taking the current K-means clustering result as a preferable clustering result; if the clustering effect evaluation index is smaller than or equal to the clustering effect evaluation index threshold, adding 1 to the K value, re-clustering all users by adopting a K-means clustering algorithm, and the like until the clustering effect evaluation index is larger than the clustering effect evaluation index threshold, and taking the last clustering result of the K-means clustering as the preferable clustering result of the K-means clustering.
Preferably, obtaining the clustering effect evaluation index based on the intra-class difference and the inter-class difference in the K-means clustering result and the difference index of the transaction data in the clustering cluster where the user to be evaluated is located in the DBSCAN clustering result includes:
Calculating the product of intra-class difference in the K-means clustering result and the difference index of transaction data in the clustering cluster where the user to be evaluated is located in the DBSCAN clustering result, and marking the product as a first index; the ratio of the difference between the first index and the class in the K-means clustering result is recorded as a first ratio;
taking a natural constant as a base, and taking the value of an exponential function with the negative first ratio as an index as a clustering effect evaluation index.
Preferably, the first risk indicator of the data to be transacted is calculated using the following formula:
wherein,for a first risk indicator of the data to be traded, < > for>For the total number of transactions of the user to be evaluated +.>For the total number of overdue money of the user to be evaluated in the course of the historical transaction, +.>For the overdue time of the i-th goods money in the history transaction process of the user to be evaluated, < +.>For the number of transactions for which the payment has not been paid so far, < >>Is an exponential function based on natural constants, < ->To adjust the parameters.
Preferably, the obtaining a change indicator of the payment time corresponding to the user to be evaluated based on the overdue duration of each overdue of the user to be evaluated and the duration between the overdue payment time and the current time includes:
any overdue repayment of the user to be evaluated in the history transaction process is performed: acquiring the time length between the overdue repayment time and the current time, and recording the time length as a first time length; taking a natural constant as a base, and taking a value of an exponential function taking the negative first duration as an index as a weight corresponding to the overdue repayment; the difference value between the overdue duration of the previous overdue of the overdue repayment and the overdue duration of the overdue repayment is recorded as a first difference value corresponding to the overdue repayment;
And taking the accumulated sum of the products of the weights corresponding to the overdue repayment and the corresponding first difference values of the users to be evaluated in the historical transaction process as a repayment time change index corresponding to the users to be evaluated.
Preferably, the following formula is used to calculate the second risk indicator of the data to be transacted:wherein (1)>For a second risk indicator of the data to be traded, < >>For the number of users in the cluster where the user to be evaluated is located in the preferred cluster result, +.>The +.f for the r-th user in the cluster where the user to be evaluated is located in the preferred cluster result>First risk index corresponding to transaction data, < >>For optimizing the number of transaction data of the r-th user in the cluster where the user to be evaluated is located in the cluster result,/L->For optimizing the difference degree of the transaction data of the r-th user and the transaction data of the user to be evaluated in the clustering cluster where the user to be evaluated is located in the clustering result, & lt/EN & gt>Is an exponential function based on natural constants; />The acquisition process of (1) is as follows: and taking the clustering distance between the transaction data of the r user and the transaction data of the user to be evaluated as the corresponding difference degree when the K-means clustering algorithm clusters for the last time.
Preferably, the determining the risk level of the payment of the user to be evaluated based on the first risk index, the second risk index and the payment time variation index of the data to be transacted includes:
Normalizing the repayment time change index, marking the difference value between the constant 1 and the normalized repayment time change index as a second difference value, and calculating the product of the second difference value, the first risk index of the data to be transacted and the second risk index of the data to be transacted to serve as the risk index of the goods to be transacted of the user to be evaluated;
judging whether the risk index of the to-be-evaluated user to be traded money is smaller than a risk index threshold value, if so, judging that the risk level of the to-be-evaluated user money is first level; and if the risk grade is greater than or equal to the second grade, judging that the risk grade of the payment of the user to be evaluated is the second grade.
The invention has at least the following beneficial effects:
1. according to the invention, transaction data in the historical transaction process of all users and personal information of all users are firstly obtained, and considering that the transaction data in the historical transaction process of the user to be evaluated can reflect the payment risk of the user to a certain extent, if the user to be evaluated has overdue payment phenomenon for a plurality of times in the historical transaction process, the overdue time length of each time is longer, and the number of transactions which are not paid yet by the user to be evaluated is more, the payment capability of the user to be evaluated is poorer, the payment risk of the user to be evaluated is higher, and if the user to be evaluated carries out transactions, the payment risk is higher; the invention obtains the first risk index based on the overdue time of each overdue of the money of the user to be evaluated, the total times of overdue of the money, the total transaction times and the transaction times which are not paid up to the present, wherein the first risk index analyzes the money risk from the behavior data layer of the user to be evaluated; considering that the money risk of the user to be evaluated can be reflected by similar transaction data of other users in the historical transaction process, if the personal information of the other users in the historical transaction process is similar to the personal information of the user to be evaluated, and the similarity degree of the transaction data of the other users to be evaluated is larger, the transaction data of the other users has higher reference value, so that the transaction data with larger similarity degree to the data to be evaluated in the historical transaction process is obtained by adopting a clustering algorithm, and a second risk index is obtained based on the transaction data with larger similarity degree, the money risk is analyzed from the behavior data level of the other users, and the correlation between the transaction data of the users is considered; considering that the self situation of the user can change continuously along with the change of time, and the repayment capability can be gradually enhanced or gradually weakened, the repayment time change index corresponding to the user to be evaluated is obtained based on the overdue time length of each overdue of the user to be evaluated and the time length between the overdue repayment time and the current time, and the repayment time change index can represent the repayment time change situation of the user to be evaluated; according to the invention, the risk of the user to be evaluated is evaluated by combining the first risk index, the second risk index and the repayment time change index, namely, the risk of the user to be evaluated is comprehensively analyzed from multiple dimensions, and the accuracy of risk evaluation of the user payment data to be evaluated is improved.
2. When transaction data with larger similarity with data to be transacted in the historical transaction process is obtained, a K-means clustering algorithm and a DBSCAN clustering algorithm are adopted to cluster all users, a preferable clustering result of the K-means clustering is obtained based on intra-class differences and inter-class differences in the K-means clustering results and difference indexes of the transaction data in the clustering clusters where the users to be evaluated are located in the DBSCAN clustering results, the similarity of the transaction data of the users in the same clustering cluster in the preferable clustering result is higher, the historical transaction data is screened based on the preferable clustering result, and the screening accuracy of the similar transaction data is higher, so that the evaluation accuracy of the subsequent goods risk of the users to be evaluated can be improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions and advantages of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are only some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a risk assessment method for payment data according to the present invention.
Detailed Description
In order to further describe the technical means and effects of the present invention for achieving the predetermined objects, a risk assessment method for payment data according to the present invention is described in detail below with reference to the accompanying drawings and preferred embodiments.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The following specifically describes a specific scheme of a risk assessment method for goods money data provided by the present invention with reference to the accompanying drawings.
An embodiment of a risk assessment method for goods money data:
the present embodiment provides a risk assessment method for payment data, as shown in fig. 1, where the risk assessment method for payment data in the present embodiment includes the following steps:
step S1, transaction data in the historical transaction process of all users and personal information of all users are obtained, wherein the transaction data comprises contracted repayment time, actual repayment time and goods money, and the personal information comprises registered funds and operation duration.
According to the method, the payment condition of the goods money in the historical transaction process of the user to be evaluated is analyzed, the risk of the goods money of the user to be evaluated in the transaction process of the historical transaction process is evaluated by combining transaction data of other users in the historical transaction process and personal information of all users, and corresponding measures are taken based on evaluation results to ensure benefits of enterprises.
In this embodiment, transaction data in the historical transaction process of all users and personal information of all users are obtained from a payment information management system of a payment company, where the transaction data includes contracted payment time, actual payment time, payment amount, and the like, and the personal information includes registered funds, operation duration, and the like. It should be noted that: the user in this embodiment refers to the company that performs the transaction. The transaction data of all users for each transaction in the historical transaction process and the personal information of all users are acquired so as to be used for evaluating the risk of the goods money data of the users to be evaluated later.
Step S2, obtaining a first risk index of data to be traded based on the overdue duration of each overdue of the money, the total number of overdue times of the money, the total number of trading times and the number of trading times which are not paid up to the present; and obtaining a repayment time change index corresponding to the user to be evaluated based on the overdue duration of each overdue of the user to be evaluated and the duration between the overdue repayment time and the current time.
In this embodiment, in step S1, transaction data in the historical transaction process of the user to be evaluated is obtained, where the transaction data in the historical transaction process of the user to be evaluated can reflect the payment risk of the user to a certain extent, if the user to be evaluated is in the historyIn the transaction process, overdue repayment occurs for a plurality of times, and the overdue duration of each time is longer, so that the repayment capability of the user to be evaluated is poorer, and the payment risk of the user to be evaluated is higher; the more the number of transactions the user to be evaluated has not paid yet, the higher the risk of payment if the user to be evaluated carries out transactions currently; based on the above, according to the overdue duration, the total overdue number, the total transaction number and the number of undelivered money to date of each overdue of the user to be evaluated, calculating a first risk index of data to be transacted, wherein the data to be transacted refers to the transaction to be currently performed with the user to be evaluated, the first risk index is obtained by analyzing the money risk from the behavior data layer of the user, and the specific expression of the first risk index of the data to be transacted is as follows:wherein (1)>For a first risk indicator of the data to be traded, < > for>For the total number of transactions of the user to be evaluated +.>For the total number of overdue money of the user to be evaluated in the course of the historical transaction, +. >For the overdue time of the i-th goods money in the history transaction process of the user to be evaluated, < +.>For the number of transactions for which the payment has not been paid so far, < >>Is an exponential function based on natural constants, < ->To adjust the parameters. Introducing adjustment parameters->In order to prevent the denominator from being 0, in this embodiment +.>The value of (2) is 0.01, and in specific applications, the practitioner can set the value according to specific situations; />Characterizing the ratio of the overdue number of the goods of the user to be evaluated to the total transaction number, wherein the larger the ratio is, the more frequent the overdue repayment of the user to be evaluated is; />The method comprises the steps of representing the total overdue duration of the money in the historical transaction process of a user to be evaluated, wherein the longer the total overdue duration of the money is, the worse the repayment capability of the user to be evaluated is; the more the number of transactions the user to be evaluated has not paid yet, the more likely the user to be evaluated has funds, and the worse the current repayment capability; when the ratio of the overdue number of the goods to the total transaction number of the users to be evaluated is larger, the total overdue duration of the goods in the historical transaction process of the users to be evaluated is longer, and the number of the transactions which are not paid yet by the users to be evaluated is larger, the repayment capacity of the users to be evaluated is poorer, and if the users to be evaluated conduct transactions, the risk of the goods is higher, namely the first risk index of the data to be transacted is larger; when the ratio of the overdue number of the goods money to the total transaction number of the user to be evaluated is smaller, the total time of overdue goods money in the historical transaction process of the user to be evaluated is shorter, the number of the transactions which are not paid yet by the user to be evaluated is smaller, the repayment capability of the user to be evaluated is higher, and if the user to be evaluated carries out transactions, the risk of goods money is lower, namely the first risk index of the data to be transacted is smaller.
Considering that the credit of the user may be continuously improved, the risk of corresponding payment data may be gradually reduced, and the closer the repayment time of the historical transaction process is to the current time, the more reference value the corresponding transaction data has, the greater attention should be paid; thus, the overdue is overdue according to the historical transaction process of the user to be evaluatedObtaining a repayment time change index corresponding to the user to be evaluated, namely, conducting overdue repayment on any time of the user to be evaluated in the historical transaction process, wherein the overdue repayment time is equal to the time between the overdue repayment time and the current time: acquiring the time length between the overdue repayment time and the current time, and recording the time length as a first time length; taking a natural constant as a base number, taking a value of an exponential function with a negative first duration as an index as a weight corresponding to the overdue repayment, recording a difference value between the overdue duration of the previous overdue of the overdue repayment and the overdue duration of the overdue repayment as a first difference value corresponding to the overdue repayment, and taking a sum of products of the weight corresponding to each overdue repayment and the corresponding first difference value in a historical transaction process of a user to be evaluated as a repayment time change index corresponding to the user to be evaluated, wherein the specific expression of the repayment time change index corresponding to the user to be evaluated is as follows: Wherein (1)>For the repayment time change index corresponding to the user to be evaluated, < > in the user to be evaluated>For the time length between the (i+1) th overdue repayment time and the current time in the history transaction process of the user to be evaluated, the user to be evaluated is added with the information>For the overdue time of the i+1st item in the history transaction process of the user to be evaluated,/for>The overdue time of the ith goods money in the history transaction process is used as the overdue time of the user to be evaluated. Note that, the duration in this embodiment is counted according to days.
If the overdue time of the user to be evaluated is shorter, the repayment capability of the user to be evaluated is gradually improved, and the payment risk index of the user to be evaluated is gradually reduced;characterizing the difference between the previous overdue duration and the later overdue duration of the user to be evaluated in the historical transaction process, and if the difference is a positive number, indicating that the overdue duration of the money of the user to be evaluated is gradually shortened, and gradually improving the repayment capability of the user to be evaluated; if the difference is negative, the payment overdue duration of the user to be evaluated is gradually increased, and the repayment capability of the user to be evaluated is gradually reduced; />Representing the weight corresponding to the (i+1) th overdue repayment in the historical transaction process of the user to be evaluated, wherein the reference meaning of the (i+1) th overdue repayment is larger as the repayment time of the historical transaction process is closer to the current time, so that the weight corresponding to the (i+1) th overdue repayment is larger; when the later overdue duration is shorter than the former overdue duration and the overdue repayment interval is shorter than the current time in the historical transaction process, the repayment capability of the user to be evaluated is gradually improved, the user to be evaluated develops in a good direction, the payment risk of the user to be evaluated is gradually reduced, and the repayment time change index corresponding to the user to be evaluated is larger than 0; when the last overdue time length is greater than or equal to the last overdue time length and the overdue repayment time interval is shorter than the current time in the historical transaction process, the repayment capability of the user to be evaluated is gradually reduced or is not changed, the repayment risk of the user to be evaluated is not reduced, and namely the repayment time change index corresponding to the user to be evaluated is less than or equal to 0.
So far, through analysis of historical transaction data of the user to be evaluated, a first risk index and a repayment time change index of the data to be evaluated are obtained.
Step S3, clustering all users by adopting a K-means clustering algorithm and a DBSCAN clustering algorithm respectively, and obtaining a preferred clustering result of the K-means clustering based on the number of users in a clustering cluster where the users to be evaluated are located in the DBSCAN clustering result, the difference degree of transaction data of each user and the data to be transacted, the intra-class difference and the inter-class difference in the K-means clustering result; the K-means clustering algorithm is based on registered funds, operation duration of each user and average amount of money in the historical transaction process, and the DBSCAN clustering algorithm is based on transaction data of the users in the historical transaction process.
In this embodiment, transaction data of all users are analyzed, transaction data similar to the data to be transacted of the user to be evaluated is screened from all users, and the risk of the goods money of the user to be evaluated is analyzed again based on the transaction data similar to the data to be transacted of the user to be evaluated in the history transaction process.
In the embodiment, a K-means clustering algorithm is adopted to cluster all users, so that transaction data similar to the data to be transacted of the users to be evaluated are screened; because the size of the clustering cluster number K cannot be determined in advance when a K-means clustering algorithm is adopted for clustering analysis, the embodiment designs a K-means clustering algorithm with a self-adaptive K value, sets an initial value of K as 2, calculates a clustering effect evaluation index after each clustering is completed, judges whether the clustering is required to be performed again based on the clustering effect evaluation index, adds 1 to the K value if the clustering effect evaluation index is required, recalculates the clustering effect evaluation index after the clustering is completed until the clustering condition is met, and otherwise, repeats the steps until the clustering condition is met.
When the K-means clustering algorithm is adopted for clustering, the clustering is carried out based on the registered funds and the operation duration of the users and the average amount of money in the historical transaction process, any two users are respectively marked as a first user and a second user, and the clustering distance between the first user and the second user is as follows:wherein (1)>For the clustering distance between the first user and the second user when the K-means clustering algorithm is adopted for clustering, the method comprises the steps of +. >For the average amount of money of the first user during the historical transaction +.>For the average amount of money of the second user during the historical transaction +.>Registering funds for the first user, +.>Registering funds for the second user, +.>For the duration of the first subscriber, +.>Is the duration of the operation of the second user.
When the average amount of the goods money, the registered funds and the operation duration of the two users in the historical transaction process are relatively close, the scales of the two companies are relatively similar, and the goods money risks are relatively similar; when the average amount of the money, the registered funds and the operating duration of the user company of the two users in the historical transaction process are different greatly, the large scale difference of the two companies is indicated, and the money risk is also different greatly.
Next, in this embodiment, the transaction data in the historical transaction process of all users is clustered by using a DBSCAN clustering algorithm, and the clustering result of the DBSCAN clustering, the intra-class difference and the inter-class difference in the K-means clustering result are combined to obtain an optimal clustering result when the K-means clustering algorithm is used for clustering. Specifically, clustering transaction data in the historical transaction process of all users by adopting a DBSCAN clustering algorithm, wherein when the DBSCAN clustering algorithm is used for clustering, any two pieces of transaction data are respectively marked as first transaction data and second transaction data, and when the DBSCAN clustering algorithm is used for clustering, the clustering distance between the clustering distances between the first transaction data and the second transaction data is as follows: Wherein (1)>For the clustering distance between the clustering distances between the first transaction data and the second transaction data when the DBSCAN clustering algorithm is adopted for clustering, the weight is +.>For the amount of money in the first transaction data, < >>For the amount of the payment in the second transaction data,for the contracted repayment time in the first transaction data,/for the transaction data>For the contracted repayment time in the second transaction data, < > in->For the duration corresponding to the first transaction data exceeding the contracted payment time,/the first transaction data>The duration exceeding the appointed repayment time corresponding to the second transaction data; the method for acquiring the duration exceeding the appointed repayment time comprises the following steps: subtracting the appointed repayment time from the actual repayment time, and taking the obtained difference value as the duration exceeding the appointed repayment time. It should be noted that: when the transaction data is stored, the data belonging to the same transaction are integrated into a sequence form for storage, and each sequence represents one transaction; selecting transaction data similar to the transaction data of the user to be evaluated from other user historical transaction data, analyzing the similar transaction data, further evaluating risks corresponding to the transaction data of the user to be evaluated, taking a clustering distance during clustering as a difference degree of the corresponding transaction data, and indicating that the transaction data are dissimilar as the difference degree is larger; k-means were polymerized The users in the cluster where the users to be evaluated are located in the class result are marked as target users, and the difference indexes of the transaction data in the cluster where the users to be evaluated are located in the DBSCAN cluster result are determined according to the difference degree between the transaction data of each target user in the cluster where the users to be evaluated are located in the DBSCAN cluster result and the transaction data of the users to be evaluated, namely: />Wherein (1)>The difference index of transaction data in a cluster where a user to be evaluated is located in a DBSCAN cluster result is +.>For the number of target users +.>The method comprises the steps of (1) selecting the number of transaction data of a b-th target user in a cluster where the data to be evaluated of the users to be evaluated in the DBSCAN cluster result is located, and (b) selecting the number of transaction data of the b-th target user in the cluster where the data to be evaluated of the users to be evaluated are located>And the difference degree of the s transaction data of the b target user and the data to be transacted in the cluster where the data to be transacted of the user to be evaluated is located in the DBSCAN cluster result.
When the difference between each transaction data and the data to be transacted in the cluster where the data to be transacted of the user to be evaluated is located in the DBSCAN cluster result is smaller, the corresponding difference degree is smaller, and the corresponding goods money risk can reflect the risk of the goods money data to be transacted of the user to be evaluated. In this embodiment, the cluster radius is set to 6 when DBSCAN clusters, and in a specific application, an implementer sets the cluster radius according to a specific situation, and a DBSCAN clustering algorithm is a known technology and will not be described in detail here.
When the clustering effect of the K-means algorithm is optimal, the intra-class difference is minimum and the inter-class difference is maximum, so the embodiment is based on the intra-class difference, the inter-class difference and the cluster where the user to be evaluated is located in the DBSCAN clustering resultJudging the clustering result of K-means clustering by using the difference index of transaction data; wherein, the calculation formulas of the intra-class difference and the inter-class difference are respectively as follows: wherein (1)>Intra-class differences ++>For the number of clusters, +.>For the number of elements in the kth cluster, +.>For the distance between the p-th element in the kth cluster and the cluster center, +.>For inter-class differences, add>Is the distance between the cluster center of the kth cluster and the cluster center of the jth cluster.
Next, in this embodiment, transaction data similar to the data to be transacted of the user to be evaluated in the historical transaction is analyzed, and then the risk of the goods money of the user to be evaluated is evaluated again, so that the more accurate the transaction data similar to the data to be transacted of the user to be evaluated is selected, the more reliable the subsequent risk of the goods money is evaluated; in the embodiment, the transaction data in the cluster where the user to be evaluated is located in the K-means clustering result is regarded as the transaction data similar to the transaction data of the user to be evaluated to be transacted, so that the clustering effect is ensured to be good enough; when the clustering effect is good, the intra-class difference is small, the inter-class difference is large, and the embodiment is based on the intra-class difference, the inter-class difference and the aggregation of the users to be evaluated in the DBSCAN clustering result Calculating a clustering effect evaluation index, namely calculating the product of intra-class difference in a K-means clustering result and a difference index of transaction data in a clustering cluster where a user to be evaluated is located in the DBSCAN clustering result, marking the product as a first index, marking the ratio of the first index to the inter-class difference in the K-means clustering result as a first ratio, taking a natural constant as a base number, taking the value of an exponential function with a negative first ratio as an index as the clustering effect evaluation index, wherein the specific expression of the clustering effect evaluation index is as follows:wherein (1)>And the clustering effect evaluation index is used. When the intra-class difference is smaller, the inter-class difference is larger, and the difference index of transaction data in a cluster where a user to be evaluated is located in a DBSCAN cluster result is smaller, the better the clustering effect is, the better the value setting of K in the K-means cluster is, namely the larger the clustering effect evaluation index is; and the clustering effect evaluation index is a clustering effect evaluation index of K-means clustering.
When the K-means algorithm is adopted for clustering, after each clustering is completed, the method is adopted to calculate the clustering effect evaluation index, and the larger the clustering effect evaluation index is, the better the clustering effect is, namely the better the setting of the value of K in the K-means clustering is, the threshold value of the clustering effect evaluation index is set When the K-means clustering algorithm is adopted for clustering, calculating a K-means clustering effect evaluation index by adopting the formula after each clustering is finished, and judging whether the K-means clustering effect evaluation index is larger than +.>If the clustering result is larger than the optimal clustering result, stopping clustering, and taking the last clustering result as the optimal clustering result; if the K-means clustering is smaller than or equal to K=K+1, clustering is carried out again until the obtained K-means clustering effect evaluation index is larger than +.>And taking the last clustering result as a preferable clustering result. In this embodiment set +.>The value of (2) is 0.7, and in a specific application, the practitioner can set according to the specific situation. The K-means clustering algorithm is a well-known technique and will not be described in detail here. According to the embodiment, a K-means clustering algorithm and a DBSCAN clustering algorithm are adopted to cluster all users, a preferable clustering result of the K-means clustering is obtained based on intra-class differences and inter-class differences in the K-means clustering results and difference indexes of transaction data in a clustering cluster where a user to be evaluated is located in the DBSCAN clustering results, the similarity degree of the transaction data of the user in the same clustering cluster in the preferable clustering results is high, the relevance among the transaction data of the user is considered, the transaction data which are similar to the user to be evaluated in terms of conditions and are high in similarity degree with the transaction data to be evaluated in the historical transaction process are selected through clustering, the transaction data in the clustering cluster where the user to be evaluated is located in the preferable clustering results are used as the transaction data which are similar to the transaction data to be evaluated, the accuracy of the similar transaction data selection is high, and further the evaluation accuracy of the commodity risk of the subsequent user to be evaluated can be improved.
Step S4, obtaining a second risk index of the data to be traded based on a first risk index of the transaction data of each user in the cluster where the user to be evaluated is located in the preferred cluster result and the difference degree of the transaction data of each user and the data to be traded; and determining the risk level of the payment of the user to be evaluated based on the first risk index, the second risk index and the repayment time change index of the data to be transacted.
In this embodiment, a preferred clustering result is obtained in step S3, transaction data similar to the transaction data to be transacted of the user to be evaluated is screened from the preferred clustering result, and the payment risk of the user under similar conditions is obtained, so as to estimate the payment risk of the user to be evaluated.
Firstly, the user to be evaluated in the preferred clustering result is in a clusterThe transaction data is taken as transaction data similar to the transaction data of the user to be evaluated to be transacted, and the payment risk of the user to be analyzed is analyzed based on the transaction data similar to the transaction data of the user to be evaluated to be transacted; if the difference degree between the transaction data in the cluster where the user to be evaluated is located in the preferred cluster result and the transaction data of the user to be evaluated is smaller, the corresponding transaction data should be given a larger attention degree, that is, the corresponding reference weight is larger, the corresponding reference weight is obtained based on the difference degree, and then the reference weight and the first user in the cluster where the user to be evaluated is located in the preferred cluster result are based on Calculating a second risk index of the data to be traded of the user to be evaluated, namely: />Wherein (1)>For a second risk indicator of the data to be traded, < >>For the number of users in the cluster where the user to be evaluated is located in the preferred cluster result, +.>The +.f for the r-th user in the cluster where the user to be evaluated is located in the preferred cluster result>First risk index corresponding to transaction data, < >>For optimizing the number of transaction data of the r-th user in the cluster where the user to be evaluated is located in the cluster result,/L->For evaluating in preferable clustering resultThe difference degree of the transaction data of the r-th user and the transaction data of the user to be evaluated in the cluster where the user is located. The obtaining process of the difference degree between the transaction data of the r user and the transaction data of the user to be evaluated in the cluster where the user to be evaluated is located in the preferred cluster result is as follows: taking the clustering distance between the transaction data of the r user and the transaction data of the user to be evaluated as the corresponding difference degree when the K-means clustering algorithm clusters for the last time; the larger the clustering distance, the more dissimilar the transaction data of the two users, i.e. the greater the degree of difference. The method comprises the steps of representing a reference weight, wherein the smaller the difference degree between transaction data of an r user and to-be-evaluated user to be transacted, the more reliable the transaction data of the r user is used for describing the risk of the transaction data of the current user, the larger the reference value, namely the larger the reference weight; when the first risk index corresponding to each piece of transaction data of each user in the cluster where the user to be evaluated is located in the preferred cluster result is larger, and the difference degree between the transaction data of each user in the cluster where the user to be evaluated is located in the preferred cluster result and the transaction data of the user to be evaluated is smaller, the risk degree of the transaction data of the user to be evaluated is higher, namely the second risk index of the transaction data of the user to be evaluated is larger; when the first risk index corresponding to each piece of transaction data of each user in the cluster where the user to be evaluated is located in the preferred cluster result is smaller, and the difference degree between the transaction data of each user in the cluster where the user to be evaluated is located in the preferred cluster result and the transaction data of the user to be evaluated is larger, the risk degree of the transaction data of the user to be evaluated is lower, namely the second risk index of the transaction data of the user to be evaluated is larger.
And finally, predicting and obtaining a second risk index of the to-be-transacted data of the user to be evaluated by analyzing the transaction data of other users.
In the embodiment, when the risk of the goods money data of the user to be evaluated is estimated, not only the historical transaction data of the user to be evaluated but also the phases of other users are consideredSimilar transaction data, combining historical transaction data of a user to be evaluated and similar transaction data of other users to be evaluated to perform risk evaluation on the payment data of the user to be evaluated, normalizing the payment time change index to be normalized to a (-1, 1) interval, wherein the normalized payment time change index can reflect the change condition of the payment capability of the user to be evaluated; therefore, according to the first risk index, the second risk index and the normalized repayment time change index of the to-be-transacted data, determining the risk index of the to-be-evaluated user to-be-transacted payment, namely, marking the difference value between the constant 1 and the normalized repayment time change index as a second difference value, calculating the product of the second difference value, the first risk index of the to-be-transacted data and the second risk index of the to-be-transacted data, and taking the product as the risk index of the to-be-evaluated user to-be-transacted payment, wherein the specific expression of the risk index of the to-be-evaluated user to-be-transacted payment is as follows: Wherein (1)>Risk indicator for the user to be evaluated for the money to be traded, < >>Is a normalized repayment time change index, < >>For a first risk indicator of the data to be traded, < > for>Is a second risk indicator for the data to be transacted.
The larger the first risk index and the second risk index are, the higher the risk of the goods money of the user to be evaluated is; the larger the normalized repayment time change index is, the more the credit of the user to be evaluated is improved, and the repayment capability is improved; when the first risk index of the data to be transacted is larger, the second risk index is larger, and the normalized repayment time change index is smaller, the risk of the goods of the user to be evaluated is higher, namely the risk index of the goods of the user to be evaluated to be transacted is larger; when the first risk index of the data to be transacted is smaller, the second risk index is smaller, and the normalized repayment time change index is larger, the risk of the goods of the user to be evaluated is lower, namely the risk index of the goods of the user to be evaluated to be transacted is smaller.
The larger the risk index of the to-be-evaluated user to be transacted goods is, the higher the risk of the goods is; setting a risk index threshold，/>The value of (2) is set according to the specific situation; judging whether the risk index of the to-be-evaluated user to be traded money is less than +. >If the risk is smaller than the first threshold, judging that the risk level of the money of the user to be evaluated is first level, namely, the risk of the money of the user to be evaluated is lower, and carrying out the transaction with the user to be evaluated; if the risk is greater than or equal to the first threshold, judging that the risk level of the money of the user to be evaluated is two-level, namely, the risk of the money of the user to be evaluated is higher, and at the moment, not suggesting to conduct the transaction with the user to be evaluated.
According to the method, transaction data in the historical transaction process of all users and personal information of all users are obtained, and considering that the transaction data in the historical transaction process of the users to be evaluated can reflect the payment risk of the users to a certain extent, if overdue payment phenomena occur for a plurality of times in the historical transaction process of the users to be evaluated, each overdue time is longer, the number of transactions which are not paid yet by the users to be evaluated is more, the payment capability of the users to be evaluated is poorer, the payment risk of the users to be evaluated is higher, and if the users to be evaluated conduct transactions with the users to be evaluated, the payment risk is higher; therefore, the embodiment obtains the first risk indicator based on the overdue time of each overdue of the money of the user to be evaluated, the total number of overdue times of the money, the total transaction number and the transaction number which are not paid so far, wherein the first risk indicator analyzes the money risk from the behavior data layer of the user to be evaluated; considering that the money risk of the user to be evaluated can be reflected by similar transaction data of other users in the historical transaction process, if the personal information of the other users in the historical transaction process is similar to the personal information of the user to be evaluated, and the similarity degree of the transaction data of the other users to be evaluated is larger, the transaction data of the other users has higher reference value, so that the embodiment adopts a clustering algorithm to acquire the transaction data with larger similarity degree to the transaction data in the historical transaction process, and obtains a second risk index based on the transaction data with larger similarity degree, and the second risk index analyzes the money risk from the aspect of the behavior data of the other users, and considers the relevance among the transaction data of the users; considering that the self situation of the user may change continuously along with the change of time, and the repayment capability may be gradually enhanced or gradually weakened, so that the repayment time change index corresponding to the user to be evaluated is obtained based on the overdue duration of each overdue of the user to be evaluated and the duration between the overdue repayment time and the current time, and the repayment time change index can represent the repayment time change situation of the user to be evaluated; according to the method and the device for evaluating the risk of the user to be evaluated, the risk of the user to be evaluated is evaluated by combining the first risk index, the second risk index and the repayment time change index, namely, the risk of the user to be evaluated is comprehensively analyzed from multiple dimensions, and accuracy of risk evaluation of the user to be evaluated is improved. When transaction data with larger similarity with data to be transacted in the historical transaction process is obtained, a K-means clustering algorithm and a DBSCAN clustering algorithm are adopted to cluster all users, a preferred clustering result of the K-means clustering is obtained based on intra-class differences and inter-class differences in the K-means clustering results and difference indexes of the transaction data in the clustering clusters where the users to be evaluated are located in the DBSCAN clustering results, the similarity of the transaction data of the users in the same clustering cluster in the preferred clustering results is higher, historical transaction data is screened based on the preferred clustering results, and the accuracy of screening of the similar transaction data is higher, so that the evaluation accuracy of the subsequent payment risk of the users to be evaluated can be improved.
Claims (7)
1. A method for risk assessment of payment data, the method comprising the steps of:
acquiring transaction data in the historical transaction process of all users and personal information of all users, wherein the transaction data comprises contracted repayment time, actual repayment time and goods money, and the personal information comprises registered funds and operation duration;
obtaining a first risk index of to-be-transacted data based on the overdue duration of each overdue of the to-be-evaluated user, the total number of overdue of the money, the total transaction number and the number of transactions which are not paid up to the present; obtaining a repayment time change index corresponding to the user to be evaluated based on the overdue duration of each overdue of the user to be evaluated and the duration between the overdue repayment time and the current time;
clustering all users by adopting a K-means clustering algorithm and a DBSCAN clustering algorithm respectively, and obtaining a preferable clustering result of the K-means clustering based on the number of users in a clustering cluster where the users to be evaluated are located in the DBSCAN clustering result, the difference degree of transaction data of each user and the data to be transacted, the intra-class difference and the inter-class difference in the K-means clustering result; the K-means clustering algorithm is based on registered funds, operation duration and average amount of money in the historical transaction process of each user, and the DBSCAN clustering algorithm is based on transaction data of the users in the historical transaction process;
Obtaining a second risk index of the data to be traded based on a first risk index of the transaction data of each user in the cluster where the user to be evaluated is located in the preferred cluster result and the difference degree of the transaction data of each user and the data to be traded; and determining the risk level of the payment of the user to be evaluated based on the first risk index, the second risk index and the repayment time change index of the data to be transacted.
2. The method for risk assessment of money data according to claim 1, wherein obtaining a preferred cluster result of K-means cluster based on the number of users in the cluster in which the users to be assessed are located in the DBSCAN cluster result, the degree of difference between the transaction data of each user and the data to be transacted, the intra-class difference and the inter-class difference in the K-means cluster result, comprises:
marking the users in the cluster where the users to be evaluated are located in the K-means cluster result as target users, and obtaining the difference index of the transaction data in the cluster where the users to be evaluated are located in the DBSCAN cluster result based on the difference degree of the transaction data and the transaction data of each target user in the cluster where the transaction data of the users to be evaluated are located in the DBSCAN cluster result;
Based on intra-class differences and inter-class differences in the K-means clustering result and difference indexes of transaction data in a clustering cluster where a user to be evaluated is located in the DBSCAN clustering result, obtaining a clustering effect evaluation index;
if the clustering effect evaluation index is larger than the clustering effect evaluation index threshold, taking the current K-means clustering result as a preferable clustering result; if the clustering effect evaluation index is smaller than or equal to the clustering effect evaluation index threshold, adding 1 to the K value, re-clustering all users by adopting a K-means clustering algorithm, and the like until the clustering effect evaluation index is larger than the clustering effect evaluation index threshold, and taking the last clustering result of the K-means clustering as the preferable clustering result of the K-means clustering.
3. The method for risk assessment of money data according to claim 2, wherein obtaining a clustering effect evaluation index based on intra-class differences and inter-class differences in K-means clustering results and difference indexes of transaction data in a cluster where a user to be assessed is located in the DBSCAN clustering results comprises:
calculating the product of intra-class difference in the K-means clustering result and the difference index of transaction data in the clustering cluster where the user to be evaluated is located in the DBSCAN clustering result, and marking the product as a first index; the ratio of the difference between the first index and the class in the K-means clustering result is recorded as a first ratio;
Taking a natural constant as a base, and taking the value of an exponential function with the negative first ratio as an index as a clustering effect evaluation index.
4. The method of claim 1, wherein the first risk indicator for the data to be traded is calculated using the formula:
wherein (1)>For a first risk indicator of the data to be traded, < > for>For the total number of transactions of the user to be evaluated +.>For the total number of overdue money of the user to be evaluated in the course of the historical transaction, +.>For the overdue time of the i-th goods money in the history transaction process of the user to be evaluated, < +.>For the number of transactions for which payment has not been paid until now,is an exponential function based on natural constants, < ->To adjust the parameters.
5. The method for risk assessment of payment data according to claim 1, wherein the obtaining a change indicator of the payment time corresponding to the user to be assessed based on the overdue duration of each overdue of the user to be assessed and the duration between the overdue payment time and the current time includes:
any overdue repayment of the user to be evaluated in the history transaction process is performed: acquiring the time length between the overdue repayment time and the current time, and recording the time length as a first time length; taking a natural constant as a base, and taking a value of an exponential function taking the negative first duration as an index as a weight corresponding to the overdue repayment; the difference value between the overdue duration of the previous overdue of the overdue repayment and the overdue duration of the overdue repayment is recorded as a first difference value corresponding to the overdue repayment; and taking the accumulated sum of the products of the weights corresponding to the overdue repayment and the corresponding first difference values of the users to be evaluated in the historical transaction process as a repayment time change index corresponding to the users to be evaluated.
6. The method of claim 1, wherein the second risk indicator for the data to be traded is calculated using the formula:wherein (1)>For a second risk indicator of the data to be traded, < >>For the number of users in the cluster where the user to be evaluated is located in the preferred cluster result, +.>The +.f for the r-th user in the cluster where the user to be evaluated is located in the preferred cluster result>First risk index corresponding to transaction data, < >>For the r user in the cluster where the user to be evaluated is located in the preferred cluster resultNumber of transaction data, < > of-> For optimizing the difference degree of the transaction data of the r-th user and the transaction data of the user to be evaluated in the clustering cluster where the user to be evaluated is located in the clustering result, & lt/EN & gt>Is an exponential function based on natural constants; />The acquisition process of (1) is as follows: and taking the clustering distance between the transaction data of the r user and the transaction data of the user to be evaluated as the corresponding difference degree when the K-means clustering algorithm clusters for the last time.
7. The method for risk assessment of payment data according to claim 1, wherein determining the risk level of the user to be assessed for the payment based on the first risk indicator, the second risk indicator, and the payment time change indicator of the data to be transacted comprises:
Normalizing the repayment time change index, marking the difference value between the constant 1 and the normalized repayment time change index as a second difference value, and calculating the product of the second difference value, the first risk index of the data to be transacted and the second risk index of the data to be transacted to serve as the risk index of the goods to be transacted of the user to be evaluated;
judging whether the risk index of the to-be-evaluated user to be traded money is smaller than a risk index threshold value, if so, judging that the risk level of the to-be-evaluated user money is first level; and if the risk grade is greater than or equal to the second grade, judging that the risk grade of the payment of the user to be evaluated is the second grade.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311579699.7A CN117291716A (en) | 2023-11-24 | 2023-11-24 | Risk assessment method for goods money data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311579699.7A CN117291716A (en) | 2023-11-24 | 2023-11-24 | Risk assessment method for goods money data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117291716A true CN117291716A (en) | 2023-12-26 |
Family
ID=89239377
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311579699.7A Pending CN117291716A (en) | 2023-11-24 | 2023-11-24 | Risk assessment method for goods money data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117291716A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107818506A (en) * | 2017-09-30 | 2018-03-20 | 上海壹账通金融科技有限公司 | Electronic installation, credit risk control method and storage medium |
| CN107993143A (en) * | 2017-11-23 | 2018-05-04 | 深圳大管加软件与技术服务有限公司 | A kind of Credit Risk Assessment method and system |
| CN109657932A (en) * | 2018-11-29 | 2019-04-19 | 平安科技（深圳）有限公司 | Business risk analysis method, device, computer equipment and storage medium |
| CN110458570A (en) * | 2019-07-02 | 2019-11-15 | 阿里巴巴集团控股有限公司 | Risk trade control and configuration method and its system |
| CN112906765A (en) * | 2021-02-01 | 2021-06-04 | 中国建设银行股份有限公司 | RBF neural network-based customer money laundering risk grading method and system |
| US20210241367A1 (en) * | 2020-01-30 | 2021-08-05 | Ajou University Industry-Academic Cooperation Foundation | Debt management capability assessment system and method of analyzing debt management capability information using the same |
| CN114820156A (en) * | 2022-03-23 | 2022-07-29 | 度小满科技(北京)有限公司 | Method, device and equipment for user risk assessment and storage medium |
-
2023
- 2023-11-24 CN CN202311579699.7A patent/CN117291716A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107818506A (en) * | 2017-09-30 | 2018-03-20 | 上海壹账通金融科技有限公司 | Electronic installation, credit risk control method and storage medium |
| CN107993143A (en) * | 2017-11-23 | 2018-05-04 | 深圳大管加软件与技术服务有限公司 | A kind of Credit Risk Assessment method and system |
| CN109657932A (en) * | 2018-11-29 | 2019-04-19 | 平安科技（深圳）有限公司 | Business risk analysis method, device, computer equipment and storage medium |
| CN110458570A (en) * | 2019-07-02 | 2019-11-15 | 阿里巴巴集团控股有限公司 | Risk trade control and configuration method and its system |
| US20210241367A1 (en) * | 2020-01-30 | 2021-08-05 | Ajou University Industry-Academic Cooperation Foundation | Debt management capability assessment system and method of analyzing debt management capability information using the same |
| CN112906765A (en) * | 2021-02-01 | 2021-06-04 | 中国建设银行股份有限公司 | RBF neural network-based customer money laundering risk grading method and system |
| CN114820156A (en) * | 2022-03-23 | 2022-07-29 | 度小满科技(北京)有限公司 | Method, device and equipment for user risk assessment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9760656B2 (en) | Methods and systems for automatic selection of classification and regression trees having preferred consistency and accuracy | |
| Bahnsen et al. | Cost sensitive credit card fraud detection using Bayes minimum risk | |
| CN110751557B (en) | Abnormal fund transaction behavior analysis method and system based on sequence model | |
| CN107657500A (en) | Stock recommends method and server | |
| Chi et al. | Bankruptcy prediction: Application of logit analysis in export credit risks | |
| Bakhach et al. | TSFDC: A trading strategy based on forecasting directional change | |
| Hamilton et al. | Revolving credit card holders: Who are they and how can they be identified? | |
| CN111915218A (en) | Financial counterfeiting identification method and system based on LSTM-CNN | |
| Hwang | Forecasting credit ratings with the varying-coefficient model | |
| Chaudhary et al. | Comparative Analysis of Entropy Weight Method and C5 Classifier for Predicting Employee Churn | |
| CN114298547A (en) | User loyalty scoring method, device, equipment and readable storage medium | |
| CN101986342A (en) | Method for arbitraging by using inherent price discrepancy of relevant finical products | |
| CN116611829B (en) | Consumption supervision system based on block chain | |
| CN116805266A (en) | Enterprise financial credit risk intelligent assessment method based on big data | |
| Sun et al. | A predicting model for accounting fraud based on ensemble learning | |
| CN117291716A (en) | Risk assessment method for goods money data | |
| Strakova et al. | Verification of earnings management in Slovak enterprises using Teoh, Welch and Wong model | |
| El Ghourabi et al. | A new financial stress index model based on support vector regression and control chart | |
| CN116738258B (en) | User key data mining method based on data analysis | |
| CN113763032B (en) | Commodity purchase intention recognition method and device | |
| CN117688455B (en) | Meta-task small sample classification method based on data quality and reinforcement learning | |
| CN116842454B (en) | Financial asset classification method and system based on support vector machine algorithm | |
| Suhanjoyo et al. | Fraud Detection in Sales of Distribution Companies Using Machine Learning | |
| Ren | Study on Intelligent Forecasting of Credit Bond Default Risk | |
| Marizal | Applied Some Mixture 2 and 3 Distribution for Daily Exchange Rate American Dollar vs Indonesian Rupiah Probability Modelling |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |