CN109886797A - A kind of batch examination & approval study of credit and optimization method - Google Patents
A kind of batch examination & approval study of credit and optimization method Download PDFInfo
- Publication number
- CN109886797A CN109886797A CN201811651868.2A CN201811651868A CN109886797A CN 109886797 A CN109886797 A CN 109886797A CN 201811651868 A CN201811651868 A CN 201811651868A CN 109886797 A CN109886797 A CN 109886797A
- Authority
- CN
- China
- Prior art keywords
- audit
- approval
- examination
- credit
- learning method
- 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
Landscapes
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
In view of the deficiencies of the prior art, the present invention discloses a kind of credit batch examination & approval learning method, first can Examination and approval personnel carry out the credit work of various amounts, and continued to optimize in supporting process.The working efficiency of approving person on the one hand can be continuously improved by optimizing, improve the speed of examination & approval；On the other hand the risk of examination & approval can be reduced as far as possible.In order to achieve the above object, the present invention is achieved by the following technical programs: examination & approval person audits at audit interface；Review process includes the following steps: that examination & approval person will screen the decision element that approval decisionmaking has a major impact from multinomial element, obtains taxis reference value, makes audit with reference to taxis reference value and determines.The technical solution of the present invention in this way, examination & approval person only audit audit element that it is chosen as decision element in review process, ignore other audit elements, reduce the process of audit.
Description
Technical field
The present invention relates to artificial intelligence fields, and in particular to a kind of credit batch examination & approval learning method.
Background technique
The fast-developing some work posts for gradually substituting various industries of artificial intelligence, financial circles are no exception.Future,
Artificial intelligence may replace artificial operational module and be broadly divided into two classes, and one kind is to follow certain step thus can be encoded into calculating
The routines of machine preview；Another kind of is not need to cope with interpersonal feelings using challenge ability or innovation ability is solved
The non-cognition of interactive or random changeable environment and emotion class work.But a process is required at everything, at this stage, we
It still needs to carry out adjustment to AI by certain methods, while can also optimize manually-operated process.
To this, application No. is the 201711270176.9 application for a patent for invention " credit risks of the small micro- loan of internet finance
Recognition methods and device " in disclose credit risk recognition methods and the device of a kind of small micro- loan of internet finance, it is described
Method is the following steps are included: obtain user credit data；User's sample is divided according to the loaning bill behavior of user；Pass through acquisition
User credit data and user's sample of division obtain an original data set；Data set is divided into training set and test
Collection, Feature Engineering is realized on training set, then reverts to these operations on test set, using the effect on test set come
Measure final performance；Selection algorithm according to the actual situation, algorithm include the two kinds of credit scorings of Logic Regression Models and xgboost
Model；Credit scoring model carries out credit scoring according to user's information.Invention user's credit audit speed is fast, manpower
It is at low cost；It comes to carry out credit audit to user from the angle of big data and machine learning, has science, while with data
Accumulation, model can continuous iteration upgrading, scoring accuracy is high.
It is analyzed and is assessed, which can bring certain side in terms of the credit risk audit of small micro- loan really
It helps.But the invention can not be applied to the high large loan of risk class, and can not assist and optimize artificial credit examination & approval work
Make.And the learning method being directed to can not be also applied in large loan scene.In addition, credit operation in the prior art
It usually requires to operate in computer end, occasionally having can be in mobile terminal operation, since the problem of interface UI causes examination & approval person to be difficult to quickly
Information is obtained, while being also easy to appear maloperation in review process.
Summary of the invention
In view of the deficiencies of the prior art, the present invention discloses a kind of credit batch examination & approval learning method, first can Examination and approval
Personnel carry out the credit work of various amounts, and continue to optimize in supporting process.It is examined by optimizing on the one hand to be continuously improved
The working efficiency for the personnel of criticizing, improves the speed of examination & approval；On the other hand the risk of examination & approval can be reduced as far as possible.
In order to achieve the above object, the present invention is achieved by the following technical programs: a kind of credit batch examination & approval study side
Method is based on credit financing auditing system, and the credit financing auditing system includes audit interface and audit database, audits data
Library obtains and is broken down into the set comprising multiple audit elements after pending project, and by this gather in audit element show
In audit interface, examination & approval person is audited at audit interface；Review process includes the following steps:
Step 1: examination & approval person will screen the decision element that approval decisionmaking has a major impact from multinomial element, and area
It is divided into positive element and negative element；
Step 2: weight assignment is carried out to each decision element；
Step 3: the weight assignment total value of positive element being subtracted to the weight assignment total value of negative element, to obtain trend
Property reference value, recommendatory audit is only made when the reference value is positive value and is determined；
Step 4: making audit with reference to taxis reference value and determine, which is determined and its forming process is recorded in audit
Database.
Further, positive element and negative element are individually positioned in positive element group and negative element group in step 1
In.In this way the element judged is convenient for distinguishing.
Preferably, in the step 2: also can specify as score value, set a total score x, belong to front greater than x/2
Element, and the more big then positive weight of score is bigger；The negative element that belongs to less than x/2, and the smaller then negative weight of score
It is bigger.
Further, after the completion of credit examination and approval project, the audit in step 4 is determined and the project final result carries out
Comparison judges the reliability that the audit determines.After making audit decision, the audit is verified by the analysis of final result and is determined
The fixed degree of reliability, and thus based on optimize audit policy.
Preferably, further include integration factor judgment step: the single audit element of judgement whether and in addition at least one audit
Element forms combination and judges element, when multiple audit elements, which form combination, judges element, recognizes the combination and judges that element is positive
Face element or negative element, while also carrying out weight assignment to it and being added to obtain in the calculating of taxis reference value.In this way
More fully and completely to the judgement of element.
Preferably, the combination obtained in integration factor judgment step is judged that element is placed on positive element group or negatively wants
It does not influence to form the audit element that the combination judges element when in plain group.I.e. if becoming one there are two negative factor combination
Positive element after combination is then placed in positive element group by positive element, if but before two negative elements be chosen as
Decision element still needs to be placed in negative element group.
Preferably, audit database judges that the audit of element is wanted to combination is likely to form when providing audit elements combination
Element is marked.Certain reminding effect is played to auditor in this way.This formed combination judge element selection be also by
What auditor manually set.
Preferably, the project final result in acquisition special time period under identical audit classification is determined as good multiple careful
Core determines and its forming process, identification in these review processes from not selected audit element, by these audit elements from
Audit is deleted in interface.Audit element is reduced in this way, improves review efficiency.
Preferably, identification is chosen as the minimum audit element of decision element frequency in these review processes, these are examined
Core element from audit interface in delete, by delete these audit elements after audit element set be loaded into audit interface in,
It is sent to other examination & approval persons to be audited, and tracks the reliability variation that judgement audit determines, if the reliability that audit determines
Do not decline, then retains time modification, it is on the contrary then recall this modification.It may insure not influence audit after review efficiency improves in this way
Accuracy.
Preferably, it repeats audit element and deletes step, be likely to result in auditing determining until deleting any audit element
Reliability decrease fixes the audit policy at this time.Audit policy is trained in this way, improves the accurate of audit policy
Property.
Further, audit policy is imported into artificial intelligence and is trained, pass through its reliability of simplation verification.Work as reality
Test sample it is enough after, artificial intelligence will carry out a large amount of sunykatuib analyses and study, on the one hand advanced optimize audit policy, another
Aspect can gradually replace artificial audit mode.
The technical solution of the present invention in this way, examination & approval person only audit its audit element conduct chosen in review process
Decision element ignores other audit elements, reduces the process of audit.Combine system regulations to careful according to the audit of examination & approval person habit
The weight assignment of core element makes the reference analysis that audit determines.And this audit policy is continued to optimize, to further mention
The working efficiency of high examination & approval person.Artificial intelligence cultivation may finally be given, realizes intelligent checks.
Specific embodiment
It in order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below will be in the embodiment of the present invention
Technical solution be clearly and completely described, it is clear that described embodiments are some of the embodiments of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
In credit batch approval process, examination & approval person is very big in face of information content to be treated of declaring, and examines each
Core case requires to handle a large amount of audit element.It is many to audit but actually during making final review decision
Element does not need to be considered.I.e. largely audit element does not interfere with final auditing result.But due to different examination & approval
The audit mode of member is different, therefore these audit elements simple cannot be treated from selecting or abandoning, but needs
Comprehensive descision is considered from the overall situation, and avoiding having ignored important audit element causes last auditing result imperfect.
To this, we devise a kind of credit batch examination & approval learning method, are based on credit financing auditing system, the finance
Credit authorization system includes audit interface and audit database, audit database obtain be broken down into after pending project include
It is multiple audit elements set, and by this gather in audit element show audit interface in, examination & approval person audit interface into
Row audit；Review process includes the following steps:
Step 1: the decision element having a major impact to examination & approval being screened from audit element, and divides into front and wants
Plain and negative element；
Step 2: weight assignment is carried out to each decision element；
Step 3: the weight assignment total value of positive element is subtracted into the weight assignment total value of negative element to obtain trend
Property reference value, recommendatory audit is only made when the reference value is positive value and is determined；
Step 4: making audit with reference to taxis reference value and determine, which is determined and its forming process is recorded in audit
Database.
In specific operation process, auditor carries out the screening of decision element in an operation interface.The operation interface
Including positive element region or negative element region.Positive element region or negative element region can be and be directly displayed at operation
In interface, such as positive element region represented with red frame, negative element region is represented with green frame.Examination & approval person can in this way
It is screened in the form of through dragging audit element to selected region.
When approving person checks the item of information for reporting and submitting material item by item, decision element can also be screened by gesture operation.
The gesture of √ marks positive element；Gesture mark negative element；Imaging material can streak portion with finger or mouse
Divide direct screenshot.
It can be ranked up by gesture in sliding, and sort and determine the significance level of element.
First method is decision element to be dragged to specified position in sliding, for example there are two elements for original, will be new
Element drag to the priority of intermediate then this element between two original elements.
This sliding is accurate but service speed is slow, so another method is the angle doing short sliding, but sliding
It can determine priority.Such as in the environment of vertical screen, the negative element area longitudinal arrangement on the right side of screen in front is to upper right sliding
Positive element, the more precipitous then priority of angle is higher when sliding to upper right.
For example, the sliding priority of the priority ratio ∠ 45° angle of the sliding element at the 60 ° of angles ∠ is higher.
It is based on weight assignment due to finally obtaining taxis reference value, it can be seen that, the weight distribution meeting of each index
Influence final result.And weight assignment how is carried out, it needs to reduce artificial subjective judgement as far as possible, is obtained by algorithm
One objective result.The method that we select herein are as follows:
Assuming that the value of each index has n, xiFor wherein i-th of value,For the average value of this n number,
For standard deviationFor xiZ- score
1. going | zi| the point of > 3, it is believed that these points are abnormal point；M (m≤n) a data are obtained after removing exceptional value, are calculated
New ordered series of numbers
Average valueThese data deduplications are obtained p (p≤n) a value, and arranged according to data from small to large by standard deviation δ
Column obtain x1, x2..., xp, corresponding frequency is ω1, ω2..., ω p.
2. n value calculates (n+1)th value before
Calculate sample skewness
Thus the assignment of weight is carried out.
In this process further include integration factor judgment step: the single audit element of judgement whether and in addition at least one
Audit element forms combination and judges element, when multiple audit elements, which form combination, judges element, recognizes the combination and judges element
For positive element or negative element, while weight assignment and being added to also is carried out to it and is obtained during taxis reference value calculates.
For example, the emergency contact that loan application people submits is son, at 20 years old son's age, this age, which is not, to sieve
The decision element of choosing, but the age of loan application people is 37 years old.Loan application man-year age and its son's age submitted differ
Too small, it is a negative decision element that two ages, which combine,.
In order to improve review efficiency, element is judged for this kind of combination, after having similar experience, audit database is being mentioned
When for audit elements combination, judges that the audit element of element is marked to combination is likely to form, examination & approval person is prompted.
After the completion of credit examination and approval project, the audit in step 4 is determined and the project final result compares, judged
The reliability that the audit determines.Project final result in acquisition special time period under identical audit classification is determined as good more
A audit determines and its forming process, and identification, from not selected audit element, these audits is wanted in these review processes
Element is deleted from audit interface.Audit element could be constantly simplified in this way, improve the working efficiency of examination & approval person.It is the generally recognized at this
It is chosen as the minimum audit element of decision element frequency in a little review processes, these audit elements are deleted from audit interface,
By delete these audit elements after audit element set be loaded into audit interface in, be sent to other examination & approval persons and examined
Core, and track the reliability variation that judgement audit determines and retain time modification, instead if the reliability that audit determines does not decline
Then recall this modification.It repeats audit element and deletes step, be likely to result in auditing determining until deleting any audit element
Determine reliability decrease, fixes the audit policy at this time.
When having certain data volume, the data finally learnt can be used for the adjustment of AI, do for the artificial intelligence in later period
Data reserve.Audit policy is imported into artificial intelligence and is trained, its reliability of simplation verification is passed through.The present invention is finally
It lays the foundation for artificial intelligent automatic audit.
The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although with reference to the foregoing embodiments
Invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each implementation
Technical solution documented by example is modified or equivalent replacement of some of the technical features；And these modification or
Replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution.
Claims (10)
1. a kind of credit batch examination & approval learning method, is based on credit financing auditing system, the credit financing auditing system includes
Audit interface and audit database, it is characterised in that: audit database is broken down into after obtaining pending project comprising multiple
Audit element set, and by this gather in audit element show audit interface in, auditor audit interface progress
Audit；Review process includes the following steps:
Step 1: examination & approval person will screen the decision element that approval decisionmaking has a major impact from multinomial element, and divide into
Positive element and negative element；
Step 2: weight assignment is carried out to each decision element；
Step 3: the weight assignment total value of positive element being subtracted to the weight assignment total value of negative element, to obtain taxis ginseng
Value is examined, recommendatory audit is only made when the reference value is positive value and is determined；
Step 4: making audit with reference to taxis reference value and determine, which is determined and its forming process is recorded in audit data
Library.
2. a kind of credit batch examination & approval learning method as described in claim 1, it is characterised in that: in the step 2: can also
To be appointed as score value, set a total score x, the positive element that belongs to greater than x/2, and the more big then positive weight of score more
Greatly；The negative element that belongs to less than x/2, and the smaller then negative weight of score is bigger.
3. a kind of credit batch examination & approval learning method as claimed in claim 1 or 2, it is characterised in that: front is wanted in step 1
Plain and negative element is individually positioned in positive element group and negative element group.
4. a kind of credit batch examination & approval learning method as described in claim 1, it is characterised in that: when credit examination and approval project is completed
Afterwards, the audit in step 4 is determined and the project final result compares, judge the reliability that the audit determines.
5. a kind of credit batch examination & approval learning method as described in claim 1, it is characterised in that: further include integration factor judgement
Step: whether the single audit element of judgement, which forms combination in addition at least one audit element, judges element, when multiple audits are wanted
Element forms combination when judge element, recognizes the combination and judges that element for front element or negative element, while also carrying out it
Weight assignment is simultaneously added in acquisition taxis reference value calculating.
6. a kind of credit batch examination & approval learning method as claimed in claim 5, it is characterised in that: by integration factor judgment step
The combination of middle acquisition, which judges not influence when element is placed in positive element group or negative element group to form the combination, judges element
Audit element.
7. a kind of credit batch examination & approval learning method as claimed in claim 2, it is characterised in that: audit database is examined in offer
When core elements combination, judge that the audit element of element is marked to combination is likely to form.
8. the optimization method that a kind of credit batch as claimed in claim 6 examines learning method, it is characterised in that: obtain specific
Project final result in period under identical audit classification is determined as good multiple audit decisions and its forming process, identification
From not selected audit element in these review processes, these audit elements are deleted from audit interface.
9. the optimization method that a kind of credit batch as claimed in claim 7 examines learning method, it is characterised in that: identification is at this
These audit elements are deleted from audit interface, will delete these by the audit element only chosen on a small quantity in a little review processes
The set of audit element after audit element is loaded into audit interface, is sent to other auditors and is audited, and tracks
Judgement audit determine reliability variation, if audit determine reliability do not decline, retain time modification, on the contrary it is then recall originally
Secondary modification.
10. the optimization method that a kind of credit batch as claimed in claim 8 examines learning method, it is characterised in that: repeat to examine
Core element deletes step, is likely to result in audit decision reliability decrease until deleting any audit element, fixing at this time should
Audit policy.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811651868.2A CN109886797A (en) | 2018-12-31 | 2018-12-31 | A kind of batch examination & approval study of credit and optimization method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811651868.2A CN109886797A (en) | 2018-12-31 | 2018-12-31 | A kind of batch examination & approval study of credit and optimization method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109886797A true CN109886797A (en) | 2019-06-14 |
Family
ID=66925523
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811651868.2A Pending CN109886797A (en) | 2018-12-31 | 2018-12-31 | A kind of batch examination & approval study of credit and optimization method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109886797A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111402034A (en) * | 2020-03-17 | 2020-07-10 | 深圳市卡牛信息科技有限公司 | Credit auditing method, device, equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102004642A (en) * | 2010-11-24 | 2011-04-06 | 中南大学 | Group collaborative solving software control technology |
| CN102034208A (en) * | 2010-12-15 | 2011-04-27 | 中国民生银行股份有限公司 | Credit operation approval data processing system and method |
| CN102034209A (en) * | 2010-12-15 | 2011-04-27 | 中国民生银行股份有限公司 | Bank credit management method and system |
| CN103246355A (en) * | 2012-02-06 | 2013-08-14 | 百度在线网络技术（北京）有限公司 | On-line input method evaluation method, system and device |
| CN103455961A (en) * | 2013-09-09 | 2013-12-18 | 丁玉兰 | Patient high risk factor nursing assessment system |
| CN106021479A (en) * | 2016-05-18 | 2016-10-12 | 广东源恒软件科技有限公司 | Project key index automatic association method and system |
| CN107844948A (en) * | 2017-10-09 | 2018-03-27 | 平安普惠企业管理有限公司 | Credit methods, device, equipment and readable storage medium storing program for executing based on credit product storehouse |
-
2018
- 2018-12-31 CN CN201811651868.2A patent/CN109886797A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102004642A (en) * | 2010-11-24 | 2011-04-06 | 中南大学 | Group collaborative solving software control technology |
| CN102034208A (en) * | 2010-12-15 | 2011-04-27 | 中国民生银行股份有限公司 | Credit operation approval data processing system and method |
| CN102034209A (en) * | 2010-12-15 | 2011-04-27 | 中国民生银行股份有限公司 | Bank credit management method and system |
| CN103246355A (en) * | 2012-02-06 | 2013-08-14 | 百度在线网络技术（北京）有限公司 | On-line input method evaluation method, system and device |
| CN103455961A (en) * | 2013-09-09 | 2013-12-18 | 丁玉兰 | Patient high risk factor nursing assessment system |
| CN106021479A (en) * | 2016-05-18 | 2016-10-12 | 广东源恒软件科技有限公司 | Project key index automatic association method and system |
| CN107844948A (en) * | 2017-10-09 | 2018-03-27 | 平安普惠企业管理有限公司 | Credit methods, device, equipment and readable storage medium storing program for executing based on credit product storehouse |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111402034A (en) * | 2020-03-17 | 2020-07-10 | 深圳市卡牛信息科技有限公司 | Credit auditing method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108898479B (en) | Credit evaluation model construction method and device | |
| TW201732662A (en) | Method and apparatus for establishing data identification model | |
| TW201734837A (en) | Multi-sampling model training method and device | |
| CN107944708A (en) | Borrow or lend money the model discrimination method, apparatus and storage medium of risk control | |
| CN107194803A (en) | P2P net loan borrower credit risk assessment device | |
| CN111461216A (en) | Case risk identification method based on machine learning | |
| CN110415111A (en) | Merge the method for logistic regression credit examination & approval with expert features based on user data | |
| CN111160473A (en) | Feature mining method and device for classified labels | |
| CN110610193A (en) | Method and device for processing labeled data | |
| CN113537807B (en) | Intelligent wind control method and equipment for enterprises | |
| CN107633030A (en) | Credit estimation method and device based on data model | |
| CN108694462B (en) | Patent retrieval result sorting method and computer-readable storage medium | |
| CN107480686B (en) | Method and device for screening machine learning characteristics | |
| CN102930495B (en) | Steganography evaluation based steganalysis method | |
| CN109063456A (en) | The safety detecting method and system of image-type identifying code | |
| CN107633455A (en) | Credit estimation method and device based on data model | |
| CN112598294A (en) | Method, device, machine readable medium and equipment for establishing scoring card model on line | |
| CN111612519B (en) | Method, device and storage medium for identifying potential customers of financial products | |
| CN108460698A (en) | Build the method and computer readable storage medium of patent value degree evaluation model | |
| CN107346515A (en) | A kind of credit card Forecasting Methodology and device by stages | |
| CN109102396A (en) | A kind of user credit ranking method, computer equipment and readable medium | |
| CN103942604B (en) | Forecasting Methodology and system based on forest discrimination model | |
| CN108629708A (en) | Build the method and computer readable storage medium of patent value degree evaluation model | |
| CN110362911A (en) | A kind of agent model selection method of Design-Oriented process | |
| CN109886797A (en) | A kind of batch examination & approval study of credit and optimization method |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190614 |