CN110400208A - The small micro- risk control model construction method of one kind and application method - Google Patents
The small micro- risk control model construction method of one kind and application method Download PDFInfo
- Publication number
- CN110400208A CN110400208A CN201810379596.9A CN201810379596A CN110400208A CN 110400208 A CN110400208 A CN 110400208A CN 201810379596 A CN201810379596 A CN 201810379596A CN 110400208 A CN110400208 A CN 110400208A
- Authority
- CN
- China
- Prior art keywords
- data
- data group
- portrait
- risk control
- control model
- 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.)
- Granted
Links
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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- 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
Abstract
The present invention relates to Financial Risk Control technical fields, more particularly to a kind of small micro- risk control model construction method and application method, wherein, the small micro- risk control model construction method of one kind, which is included in initial client portrait library, to be screened to form a sample database by preordering method, and forms determining data group according to the sample database；It calculates to form export data group data according to the determining data group based on EM-tool algorithm；One deep learning group data are formed according to the export data group, the determining data group based on vector machine algorithm.
Description
Technical field
The present invention relates to Financial Risk Control technical fields, more particularly to a kind of small micro- risk control model construction method
And application method.
Background technique
The starting point of the small micro- finance of China Businessization is in 2005, undertaken derived from World Bank's initiation, State Development Bank,
German IPC company provides [the small wechat credit item mesh of State Development Bank] that skill is helped.The project will be commercialized small wechat for the first time and borrow
Air control technology introduces China, supports 18 municipal commercial banks in total under the project and Rural Commercial Bank carries out small wechat
Loan business and the small micro- credit technique of IPC is propagated, leader Taizhou bank, concessionaire bank, Chongqing bank for example including the field,
Kweiyang bank, Anhui Maanshan agriculture firm etc..From 2008, a large amount of municipal commercial bank, Rural Commercial Bank, rural area letter
The small wechat of IPC is introduced and propagated with association, little Dai company and borrows air control technology, which becomes each financial institution and carry out small wechat
The core air control technology of loan business.
The core that the small wechat of existing IPC borrows air control technology is that restore those by " crosscheck " technology non-just without report
The financial statement at the small difference quotient family of rule is manually examined credit and can be examined quickly provide a loan by what is authorized extensively.The small gentle breeze control of IPC
The first loan repayment capacity that technology is capable of accurate calculation client avoids bull from being in debt, and the refund wish of assessment client, thus real
Now by credit risk control before loan.By taking Taizhou bank practices the practice of IPC credit technique as an example, its small micro- industry over more than 10 years
The air control that business keeps below 1% for a long time is horizontal, and the validity of the small micro- credit technique of IPC is demonstrated with good asset quality.
The small wechat of IPC borrows air control technology and some significant drawbacks is exposed in many years extension process, such as: one, it is tight
" Almightiness type " customer manager is relied on again, and " Almightiness type " customer manager usually requires the longer training period, usual customer manager's
The training introduction time is 3 months, and basic grasp is 6 months, carries out all kinds of business comprehensively and needs 12 months, this is for wishing quick
Form the technological reserve phase for needing to grow very much for the mechanism of business scale.At the same time, when customer manager can independently lead
When team, then corresponding audit technology is passed on by the formation of " master worker trains an apprentice ", but one " master worker " is difficult to grasp comprehensively
All audit technologies or skill, so the technology or skill of succession are constantly shunk, Zhou Erfu in the technology succession in later period
Begin constantly to form a vicious circle；Two, operation wind of the small gentle breeze control technology of IPC centered on customer manager, in credit process
Danger is higher, and one people of customer manager completes marketing, investigation, tabulation, post-loan management.Slightly experienced customer manager and client's tradition
Data falsification and the threshold of data are low, and how to take precautions against the moral hazard prevention of small micro- customer manager is often that each financial institution answers
With the pain spot of IPC technology；Three, meeting examination and approval system height dependence of examining credit under the small gentle breeze control technology line of IPC has abundant air control experience
Believe the personnel of examining, examines low efficiency.The meeting power of examination and approval of examining credit under usual IPC line is come true in strict accordance with the respective permission combination of approver
It is fixed, therefore the personnel for usually requiring the higher power of examination and approval for often resulting in branch participate in, or due to examining and investigating point
From principle, cause to have the approving person of investigation permission cannot examine relevant loan, these all cause examination & approval
It is inefficient.
Summary of the invention
In view of the above-mentioned problems, the embodiment of the invention provides a kind of small micro- risk control model construction method and application sides
Method, it is intended to improve review efficiency, reduce risk, reduce cost of labor.
On the one hand, the disclosure provides a kind of small micro- risk control model construction method, in which: including,
It screens to form a sample database by preordering method in initial client portrait library, and according to the sample database
It is formed and determines data group；
It calculates to form export data group data according to the determining data group based on EM-tool algorithm；
One deep learning group data are formed according to the export data group, the determining data group based on vector machine algorithm.
Preferably, above-mentioned small micro- risk control model construction method, in which: including Yu Suoshu initial client portrait library
In screen to form a sample database by preordering method, and determining data group is formed according to the sample database and includes:
Step S11, M first kind index item, N number of second class index item, the first kind index item, described second are set
Class index item is configured with Q option value；According to the option value of the first kind index item, the option value of the second class index item
Form initial client portrait library and portrait combination；
Step S12, any customer portrait data in initial client portrait library are read, and to the current client
Representation data is analyzed and determined to form a judging result；
Step S13, it is formed according to the judging result and updates the initial portrait library, reading is got rid of according to the judging result
The portrait number of combinations removed executes step S12 not less than the state of predetermined value in the portrait number of combinations forgone；Instead
It, forms the sample database according to the updated initial portrait library.
Preferably, the above-mentioned small micro- risk control model construction method of one kind, in which: in Yu Suoshu step S12, read
Any customer portrait data in initial client portrait library, and the current customer portrait data are analyzed and determined with shape
It is specifically included at a judging result:
Step S121, it is judged as the state of refusal in presently described customer portrait data, gets rid of lower than presently described visitor
The combination pond of family representation data；
Step S122, it is judged as the state of approval in presently described customer portrait data, receives to be higher than presently described visitor
The combination pond of family representation data.
Preferably, the above-mentioned small micro- risk control model construction method of one kind, based on EM-tool algorithm according to the determination
Data group calculate to be formed export data group data specifically include:
The matched percent of pass of each option value of group zygonema in the determining data group with each representation data is read,
Percent of pass based on option value described in each obtains the percent of pass of presently described representation data group zygonema；
Sequence processing is done to form the export data group data the percent of pass of each portrait group zygonema.
Preferably, the above-mentioned small micro- risk control model construction method of one kind, based on vector machine algorithm according to the export
Data group, the determining data group form a deep learning group data and include:
The determining data group data, the export data group data are read, weight is done to the export data group data
Processing is to form weight export data group data；
Vector machine processing is done according to the determining data group data, weight export data group data to form the depth
Habit group data.
On the other hand, the present invention is providing a kind of application method of small micro- risk control model, wherein including based on above-mentioned
The risk control model that described in any item small micro- risk control model construction methods of one kind are formed, further includes:
The representation data to be evaluated for reading user is obtained according to the representation data to be evaluated in conjunction with risk control model calculating
Take highest that can borrow amount；
Ratify the loan of active user in the state that the highest can borrow target loan limit of the amount not less than user
Request；
In the state that the highest can borrow target loan limit of the amount less than user, amount can be borrowed according to the highest
It is formed and suggests loan limit.
Compared with the prior art, the advantages of the present invention are as follows:
Traditional modeling approach that expert model is constructed by subjective setting index item and manual allocation weight has been abandoned,
In order to substantially provide expert model precision, constructs first or update initial client portrait library, the expert for the modeling that lets on is based on just
Beginning customer portrait library carries out 0 or 1 and judges to form determining data group.Secondly, being obtained based on determining data group by EM-tool algorithm
It is surplus to determine that data group data, export data group data are completed finally by deep learning for the export data group data for having taken 70%
Remaining and 30% calculate, it is on the one hand that the subjectivity of expert judging is near minimum, to greatly improve the precision of modeling, on the other hand without
It needs a large amount of veteran letters to examine personnel to examine, substantially increases the efficiency of examination & approval while reducing the cost of examination & approval.
Detailed description of the invention
Fig. 1 is the flow chart of the small micro- risk control model construction method of one of the embodiment of the present invention；
Fig. 2 is the flow chart of the small micro- risk control model construction method of one of the embodiment of the present invention；
Fig. 3 is the flow chart of the small micro- risk control model construction method of one of the embodiment of the present invention；
Fig. 4 is the flow chart of the small micro- risk control model construction method of one of the embodiment of the present invention；
Fig. 5 is the flow chart of the small micro- risk control model construction method of one of the embodiment of the present invention；
Fig. 6 is the flow chart of one of the embodiment of the present invention application method of small micro- risk control model.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention rather than limiting the invention.It also should be noted that in order to just
Only the parts related to the present invention are shown in description, attached drawing rather than entire infrastructure.
Embodiment one
Fig. 1 is the flow chart of the small micro- risk control model construction method of one of embodiment of the present invention one, the present embodiment
It is applicable to the situation of any mobile terminal.This method can be executed by server, which can be using software and/or hard
The mode of part is realized.As shown in Figure 1, for the process of the small micro- risk control model construction method of one of the embodiment of the present invention one
Figure, this method specifically include:
Step S1, it screens to form a sample database by preordering method in initial client portrait library, and according to the sample
Database forms determining data group；It specifically includes:
As shown in Fig. 2, step S11, setting M first kind index item, N number of second class index item, the first kind index
Item, the second class index item are configured with Q option value；Referred to according to the option value of the first kind index item, second class
The option value for marking item forms initial client portrait library and portrait combination；Wherein first kind index item can be qualitative index item, such as
Marital status, occupancy etc., the second class index item can be quantitative target item, such as debt ratio, moon disposable income data, loan
The data such as the money installment reimbursement amount of money, the sum of M and N are at least 10, and wherein M is natural number, N is natural number；The M first kind
Q option value is each equipped in index item or N number of second class index item, the value range of Q can be 2~4, each customer portrait
First kind index item, the second class index item in data are corresponding with an option value, i.e., each customer portrait data are by first
The option value of class index item and the second class index item is formed.
Step S12, any customer portrait data in initial client portrait library are read, and to the current client
Representation data is analyzed and determined to form a judging result；It wherein include M first kind index in each customer portrait data
Result, the option value of N number of second class index item of item, the judging result can be completed by expert, can also be completed by calculator, herein
It is not particularly limited.Wherein as shown in figure 3, specifically including:
Step S121, it is judged as the state of refusal in presently described customer portrait data, forgoes lower than presently described visitor
The combination pond of family representation data；Lower than all customer portrait data in the combination pond of existing customer representation data, pass through
Probability be respectively less than existing customer representation data by probability, so directly forgo.
Step S122, it is judged as the state of approval in presently described customer portrait data, receives to be higher than presently described visitor
The combination pond of family representation data.Higher than all customer portrait data in the combination pond of presently described customer portrait data, lead to
The probability crossed be not less than existing customer representation data by probability, so can directly receive.Execute step S121 or step
After rapid S122, a customer portrait group zygonema can get, initial client can be further reduced according to the customer portrait group zygonema and drawn
As the customer portrait quantity in library.
Step S13, it is formed according to the judging result and updates the initial portrait library, reading is got rid of according to the judging result
The portrait number of combinations removed executes step S12 not less than the state of predetermined value in the portrait number of combinations forgone；Instead
It, forms the sample database according to the updated initial portrait library.Wherein predetermined value is can be 5~10, i.e.,
It can be only forgone after judgement in the state that 5~10 portraits combine every time and then stop judging.Wherein, it forgoes after judgement every time
The portrait combination quantity be greater than 10 in the state of, can determine that as there are still bulk redundancy numbers in current initial portrait library
According to needing to be done according to existing customer portrait group zygonema at this time and further forgo processing.5 can be only forgone after judging every time~
Then stopping judging in the state of 10 portrait combinations, there are still the data for partially needing to forgo in initial portrait library at this time, but
Be forgo it is relatively inefficient, so stop judgement.
Step S2, it calculates to form export data group data according to the determining data group based on EM-tool algorithm；Specific packet
It includes: as shown in figure 4, step S21 reads each option value of group zygonema in the determining data group with each representation data
Matched percent of pass, the percent of pass based on option value described in each obtain the percent of pass of presently described representation data group zygonema；
Step S22, sequence processing is done to form the export data group the percent of pass of each portrait group zygonema
Data.
Step S3, a deep learning is formed according to the export data group, the determining data group based on vector machine algorithm
Group data.Specifically, as shown in Figure 5, comprising:
Step S31, the determining data group data, the export data group data are read, to the export data group number
According to weight processing is done to form weight export data group data；The weight for exporting data group data can be 0.2~1.5.
Step S32, vector machine processing is done to be formed according to the determining data group data, weight export data group data
State deep learning group data.
Enumerate a specific embodiment, in the embodiment, have artificial judgment formed judging result, machine also it is achievable this
Judgement, principle is similar to artificial judgment principle, herein only by taking artificial judgment as an example, filters out corresponding expert person first,
In small micro- risk control model building, believe that examining expert participates in artificial judgment modeling, believes for every and examines expert extremely by 6
It examines credit and can undergo under the line less with 5 years.The weight for determining each expert first determines final expert using 3 indexs
Weight, calculates the identifiability F1 of each expert's flag data corresponding data mode, the stability F2 of data, and expert's business is comprehensive
Tri- finger target values of conjunction ability F3, wherein F1 and F2 is objective indicator, can be directly calculated by data；F3 is subjective
Index is tested and assessed using the method manually evaluated and tested, chooses multiple evaluation people, and each evaluation per capita more understands each expert, is led to
Excessive people obtains the corresponding expert's service integration ability of the expert to the evaluation of each expert, and the importance of three indexs is the same,
It is configured according to equal weight.
Obtaining mode identifiability F1, mode identifiability F1 are the recognizable degree of expert's mark data, i.e., specially from this
Learn its data rule in the data that family marked, the boundary property of expert's labeled data is obtained using different algorithms, can be used
The variance of multiple XGBOOST algorithm accuracy is measured, and the bigger quality of data for illustrating expert label of variance is poor, algorithm
Habit energy variation is larger, and variance is the smaller the better, the variance of every expert is calculated with this, and be standardized to it, so that it may
Obtain the F1 index of every expert.
The stability F2 of expert is obtained, the stability F2 of expert is a certain expert front and back at least twice to same data markers
Consistency, the stability of expert is higher, illustrates that it is more deep to the understanding of standard object, and flag data is more stable, number
It is higher according to quality confidence level.
Expert's service integration ability F3 is obtained, expert's service integration ability F3 refers to the service integration ability of expert, at least
It include: the professional skill of expert, the academic title of expert, the educational background of expert, the length of service of expert, personality level of expert etc. side
Face.The service integration ability of expert is higher, illustrates that the quality of expert data is relatively high, the quality of data is more credible.
An expert is formed based on mode identifiability F1, the stability F2 of expert, expert service integration ability F3 to calculate
Weight.
Each expert reads any customer portrait data in initial client portrait library, and wherein customer portrait is
34992, and the current customer portrait data are analyzed and determined to form a judging result；
It include the option of the result of M first kind index item, N number of second class index item in each customer portrait data
Value, expert obtain customer portrait data information from customer portrait library, read each option value in each customer portrait data,
And according to option value judge the customer portrait data corresponding to loan requests can be passed through, when judging result be 0 state
Under, then it represents that refuse the corresponding loan requests of customer portrait data, in the state that judging result is 1, then it represents that allow this
The corresponding loan requests of customer portrait data.In the state that judging result is 0, using existing customer representation data as Threshold extent
Value, it is all lower than the customer portrait data of the threshold limit value to forgo.Eventually forming one has expert decision-making to obtain determining data
Group, 34992 customer datas are drawn a portrait in combination, and expert needs to judge 1000-1500 portraits combinations.Continue through EM-
Tool algorithm calculates to form export data group data according to the determining data group；Wherein computable export data combination is general
It is 24000.Deep learning group is obtained using XGBOOST algorithm based on determining data group data, the export data group data
Data, deep learning group data are about 10000.In above-mentioned case study on implementation, data group data, export data group number are determined
According to, deep learning group data layout, the wherein accuracy rate of deep learning group data, at least up to 85%.
In the above method, abandon traditional index item and manual allocation weight are set to construct expert model by subjectivity
Modeling approach construct customer portrait library first to substantially provide expert model precision, the expert for the modeling that then lets on
0 or 1 is only carried out to judge to form determining data group.Secondly, obtaining 70% by EM-tool algorithm based on determining data group
Export data group data, finally by deep learning determine data group data, export data group data complete residue 30% count
It calculates, it is on the one hand that the subjectivity of expert judging is near minimum, so that the precision of modeling is greatly improved, on the other hand without a large amount of warps
It tests the careful personnel of letter abundant to examine, substantially increases the efficiency of examination & approval while reducing the cost of examination & approval.
Embodiment two
Based on the above embodiment one, the present embodiment is in a kind of open application method of small micro- risk control model, such as Fig. 6 institute
The flow chart of the application method of the small micro- risk control model of one kind shown, including described in any one that provides based on the above embodiment
The risk control model that is formed of the small micro- risk control model construction method of one kind, further includes:
Step S201, the representation data to be evaluated for reading user, according to the representation data to be evaluated in conjunction with the risk control
Model, which calculates acquisition highest, can borrow amount；
Step S202, Yu Suoshu highest can borrow amount not less than the current use of approval in the state of the target loan limit of user
The loan requests at family；
Step S203, Yu Suoshu highest can borrow amount less than user target loan limit in the state of, according to it is described most
Height, which can borrow amount and be formed, suggests loan limit.
Enumerate a specific embodiment
Assuming that all directions of a certain user are designated as X0, wherein X0=[x1；x2；x3；x4；x5；x6；x7；x8；x9；
X10], first kind index item is G0, wherein G0=[x1；x2；x3；x4；；x6；；； x9；], the second index item is H0,
H0=[；；；；x5；；x7；x8；；x10]；Using G0 as querying condition, filtered out from risk control model all full
The record of sufficient G0 combination, forms the matrix E1 for the condition that meets, and then pushes away Shen monetary allowance volume by the way that 4 boundary thresholds in HO are counter, so
Take the minimum value of four boundary values can monetary allowance volume as highest afterwards.Amount can be borrowed in the highest to provide a loan not less than the target of user
Ratify the demand for loan of active user in the state of amount, Yu Suoshu highest can borrow target loan limit of the amount less than user
Under state, it can borrow amount according to the highest and be formed and suggest loan limit.
Note that the above is only a better embodiment of the present invention and the applied technical principle.It will be appreciated by those skilled in the art that
The invention is not limited to the specific embodiments described herein, be able to carry out for a person skilled in the art it is various it is apparent variation,
It readjusts and substitutes without departing from protection scope of the present invention.Therefore, although being carried out by above embodiments to the present invention
It is described in further detail, but the present invention is not limited to the above embodiments only, without departing from the inventive concept, also
It may include more other equivalent embodiments, and the scope of the invention is determined by the scope of the appended claims.
Claims (6)
1. a kind of small micro- risk control model construction method, it is characterised in that: including,
It screens to form a sample database by preordering method in initial client portrait library, and is formed according to the sample database
Determine data group；
It calculates to form export data group data according to the determining data group based on EM-tool algorithm；
One deep learning group data are formed according to the export data group, the determining data group based on vector machine algorithm.
2. small micro- risk control model construction method according to claim 1, it is characterised in that: including in described initial
It screens to form a sample database by preordering method in customer portrait library, and determining data group is formed according to the sample database
Include:
Step S11, M first kind index item, N number of second class index item are set, and the first kind index item, second class refer to
It marks item and is configured with Q option value；It is formed according to the option value of the option value of the first kind index item, the second class index item
Initial client portrait library and portrait combination；
Step S12, any customer portrait data in initial client portrait library are read, and to the current customer portrait
Data are analyzed and determined to form a judging result；
Step S13, it is formed according to the judging result and updates the initial portrait library, what reading was forgone according to the judging result
Portrait number of combinations executes step S12 not less than the state of predetermined value in the portrait number of combinations forgone；Conversely, root
The sample database is formed according to the updated initial portrait library.
3. the small micro- risk control model construction method of one kind according to claim 2, it is characterised in that: Yu Suoshu step
In S12, any customer portrait data in initial client portrait library are read, and to the current customer portrait data point
Analysis judges to form a judging result and specifically include:
Step S121, it is judged as the state of refusal in presently described customer portrait data, gets rid of and is drawn lower than presently described client
As the combination pond of data；
Step S122, it is judged as the state of approval in presently described customer portrait data, receives to be higher than presently described client picture
As the combination pond of data.
4. the small micro- risk control model construction method of one kind according to claim 2, it is characterised in that: be based on EM-tool
Algorithm calculated according to the determining data group to be formed export data group data specifically include:
The matched percent of pass of each option value of group zygonema in the determining data group with each representation data is read, is based on
The percent of pass of each option value obtains the percent of pass of presently described representation data group zygonema；
Sequence processing is done to form the export data group data the percent of pass of each portrait group zygonema.
5. the small micro- risk control model construction method of one kind according to claim 2, it is characterised in that: calculated based on vector machine
Method forms a deep learning group data according to the export data group, the determining data group
The determining data group data, the export data group data are read, weight processing is done to the export data group data
Data group data is exported to form a weight；
Vector machine processing is done according to the determining data group data, weight export data group data to form the deep learning group
Data.
6. a kind of application method of small micro- risk control model, which is characterized in that including any based on the claims 1~5
The risk control model that the small micro- risk control model construction method of one kind described in is formed, further includes:
The representation data to be evaluated for reading user is calculated in conjunction with the risk control model according to the representation data to be evaluated and is obtained most
Height can borrow amount；
Ratify the loan requests of active user in the state that the highest can borrow target loan limit of the amount not less than user；
In the state that the highest can borrow target loan limit of the amount less than user, amount can be borrowed according to the highest and formed
It is recommended that loan limit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810379596.9A CN110400208B (en) | 2018-04-25 | 2018-04-25 | Small and micro risk control model construction method and application method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810379596.9A CN110400208B (en) | 2018-04-25 | 2018-04-25 | Small and micro risk control model construction method and application method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110400208A true CN110400208A (en) | 2019-11-01 |
| CN110400208B CN110400208B (en) | 2021-03-12 |
Family
ID=68319717
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810379596.9A Active CN110400208B (en) | 2018-04-25 | 2018-04-25 | Small and micro risk control model construction method and application method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110400208B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102135979A (en) * | 2010-12-08 | 2011-07-27 | 华为技术有限公司 | Data cleaning method and device |
| US20160196587A1 (en) * | 2002-02-07 | 2016-07-07 | Asset Reliance, Inc. | Predictive modeling system applied to contextual commerce |
| CN106447490A (en) * | 2016-09-26 | 2017-02-22 | 广州速鸿信息科技有限公司 | Credit investigation application method based on user figures |
| CN107025596A (en) * | 2016-02-01 | 2017-08-08 | 腾讯科技（深圳）有限公司 | A kind of methods of risk assessment and system |
| CN107798390A (en) * | 2017-11-22 | 2018-03-13 | 阿里巴巴集团控股有限公司 | A kind of training method of machine learning model, device and electronic equipment |
-
2018
- 2018-04-25 CN CN201810379596.9A patent/CN110400208B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160196587A1 (en) * | 2002-02-07 | 2016-07-07 | Asset Reliance, Inc. | Predictive modeling system applied to contextual commerce |
| CN102135979A (en) * | 2010-12-08 | 2011-07-27 | 华为技术有限公司 | Data cleaning method and device |
| CN107025596A (en) * | 2016-02-01 | 2017-08-08 | 腾讯科技（深圳）有限公司 | A kind of methods of risk assessment and system |
| CN106447490A (en) * | 2016-09-26 | 2017-02-22 | 广州速鸿信息科技有限公司 | Credit investigation application method based on user figures |
| CN107798390A (en) * | 2017-11-22 | 2018-03-13 | 阿里巴巴集团控股有限公司 | A kind of training method of machine learning model, device and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110400208B (en) | 2021-03-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109360084A (en) | Appraisal procedure and device, storage medium, the computer equipment of reference default risk | |
| CN106651573A (en) | Business data processing method and apparatus | |
| Masduki et al. | How can quality regional spending reduce poverty and improve human development index? | |
| Nadhir et al. | The effect of audit quality and degree of international Financial Reporting Standards (IFRS) convergence on the accrual earnings management in ASEAN countries | |
| CN106910016A (en) | Vocational skills appraisal procedure and system | |
| CN109118038A (en) | The risk rating method and device of block chain project | |
| Costeiu et al. | Bridging the banking sector with the real economy: a financial stability perspective | |
| Dziwok | New approach to operational risk measurement in banks | |
| CN110400208A (en) | The small micro- risk control model construction method of one kind and application method | |
| Li et al. | Do remittances foster financial inclusion in Mexico | |
| Ibiyemi et al. | Critical issues in economic risks consideration by commercial property investors and valuers in Nigeria: the case of Lagos | |
| Davis et al. | Method: The New Zealand Socio‐economic Index of Occupational Status: methodological revision and imputation for missing data | |
| Alfeus et al. | An empirical study of the option pricing formula with the underlying banned from short sell | |
| Liu et al. | Pro-growth inefficiency: Rents and moral hazard in infrastructure contests in China | |
| Yang et al. | Risk evaluation of highway engineering project based on the fuzzy-AHP | |
| Shah et al. | Development and Validation of Measurement Scale for Perceived Ease of Doing Business | |
| Martins et al. | The determinants of macroeconomic forecasts and the stability and and growth pact | |
| Akin | Determination of sovereign credit rating model for European countries | |
| Leandro | An internal fraud model for operational losses: an application to evaluate data integration techniques in operational risk management in financial institutions | |
| Demartini et al. | Audit quality | |
| Chen et al. | Research on the Impact of Relationship Networks on Farmers’ Formal Credit Constraints: The Case of China | |
| Mba | An Examination of Discounted Cash Flow Valuation Model as it relates to Property Investments in Nigeria | |
| McConnell | Prime loss: A case study in operational risk. | |
| Miankhel et al. | Post Global Financial Crisis Export inefficiencies of India: A baseline for post COVID analyses | |
| O’Donoghue et al. | The spatial distribution of household disposable income |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |