CN109166034A - A kind of Risk Forecast Method and system - Google Patents
A kind of Risk Forecast Method and system Download PDFInfo
- Publication number
- CN109166034A CN109166034A CN201811085132.3A CN201811085132A CN109166034A CN 109166034 A CN109166034 A CN 109166034A CN 201811085132 A CN201811085132 A CN 201811085132A CN 109166034 A CN109166034 A CN 109166034A
- Authority
- CN
- China
- Prior art keywords
- loan application
- period
- prediction
- predicted
- application amount
- 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
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
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of Risk Forecast Method and systems, comprising: using the time as granularity, obtains the data information of loan application；According to the data information of the loan application, creation obtains prediction model；The loan application amount that each period in predicted time is treated based on the prediction model is predicted, the prediction result of the loan application amount of each period is obtained；According to the prediction result, risk assessment is carried out to the loan application amount of each period, obtains risk evaluation result.The purpose for improving forecasting efficiency and precision of prediction is realized through the invention.
Description
Technical field
The present invention relates to financial technology fields, more particularly to a kind of Risk Forecast Method and system.
Background technique
Along with the development of Internet technology, banking internet can provide convenient and efficient client's body to client
It tests, but incident risk high concentration also becomes main problem.
For example, on bank's line in the schedule operation of loan system business, if without fraud or promotion activity etc. is concentrated
Event occurs, and loan system will not be affected during operation on bank's line.But events such as loan fraud if it exists
Meeting so that system operating be affected, for example, the efficiency of data processing can be reduced, will affect the processing of normal loan application business
Process etc..And the existing risk profile to loan system business on bank's line is mainly based upon the audit of staff and divides
Analysis, can make forecasting efficiency low and precision of prediction declines.
Summary of the invention
Be directed to the above problem, the present invention provides a kind of Risk Forecast Method and system, realize improve forecasting efficiency and
The purpose of precision of prediction.
To achieve the goals above, the present invention provides the following technical scheme that
A kind of Risk Forecast Method, comprising:
Using the time as granularity, the data information of loan application is obtained；
According to the data information of the loan application, creation obtains prediction model；
The loan application amount that each period in predicted time is treated based on the prediction model is predicted, described in acquisition
The prediction result of the loan application amount of each period；
According to the prediction result, risk assessment is carried out to the loan application amount of each period, obtains risk assessment
As a result.
Optionally, further includes:
If the risk evaluation result shows that the loan application amount of period to be predicted has exception, it is described pre- to calculate acquisition
The difference of the predicted quantity of the loan application amount for the prediction period that the loan application amount and the prediction model for surveying the period obtain
Value；
According to the difference, fusing rank is determined；
According to the fusing rank, fusing processing is carried out to period to be predicted corresponding loan application, is realized to institute
State the risk response of loan application.
Optionally, the loan application amount that each period in predicted time is treated based on the prediction model is carried out pre-
It surveys, obtains the prediction result of the loan application amount of each period, comprising:
The loan application amount in predicted time is treated based on the prediction model to be predicted, initial prediction is obtained；
Obtain the loan application amount actual numerical value of the first period collection in the time to be predicted；
Based on the loan application amount actual numerical value, the initial prediction is adjusted, is obtained in the time to be predicted
The second period collection each period loan application amount prediction result, wherein the first period collection and it is described second when
Section collection meets preset time corresponding relationship.
Optionally, it is described be based on the loan application amount actual numerical value, the initial prediction is adjusted, obtain to
The prediction result of the loan application amount of each period of the second period collection in predicted time, comprising:
Obtain the ratio that each period loan application amount accounts for loan application total amount；
Loan application amount actual numerical value and the corresponding ratio of the first period collection based on the first period collection are calculated and are obtained
The calculated value of loan application total amount；
According to the calculated value of the loan application total amount, the predicted value for treating the loan application total amount in predicted time is carried out
Adjustment obtains target Prediction of Total value；
The ratio of loan application total amount is accounted for according to the second period collection, is calculated and is obtained the second period collection in the time to be predicted
The prediction result of the loan application amount of each period.
Optionally, after calculating the calculated value for obtaining loan application total amount, this method further include:
Calculate the calculated value of the loan application total amount and the predicted value of the loan application total amount in the time to be predicted
Between absolute difference；
Judge whether the absolute difference meets default adjustment threshold value, if it is, according to the loan application total amount
Calculated value, the predicted value for treating the loan application total amount in predicted time are adjusted, and obtain target Prediction of Total value.
Optionally, the data information according to the loan application, creation obtain prediction model, comprising:
According to the data information of the loan application, loan application amount time series is generated；
According to the loan application amount time series, the first prediction model is created；
Information extraction is carried out to the data information of the loan application, obtains influence factor parameter information；
Based on the influence factor parameter information, the second prediction model is created；
It calculates separately and obtains first prediction model and the corresponding forecasting effective measure of second prediction model
Based on the forecasting effective measure, target prediction is determined in first prediction model and second prediction model
Model.
Optionally, when first prediction model is arma modeling, then described according to the loan application amount time sequence
Column create the first prediction model, comprising:
According to the loan application amount time series, the auto-correlation coefficient and partial correlation coefficient for obtaining the sequence are calculated；
According to the auto-correlation coefficient and partial correlation coefficient, the order of arma modeling is determined；
Using the arma modeling for having determined that order, the parameter of the arma modeling is obtained；
The arma modeling is verified according to the parameter, and is fitted based on verification result, several are obtained
Model of fit determines target arma modeling in the model of fit.
Optionally, when second prediction model is BP neural network model, then described to be joined based on the influence factor
Number information, creates the second prediction model, comprising:
According to the influence factor parameter information, the input sample collection of BP neural network model is determined；
According to practical loan application quantity, the output sample set of BP neural network model is established；
According to the input sample collection and the output sample set, establishes and obtain the BP neural network model；
Parameter adjustment is carried out to the BP neural network model, and carries out models fitting according to parameter adjusted, is determined
Target BP neural network model.
A kind of Risk Forecast System, comprising:
Acquiring unit, for obtaining the data information of loan application using the time as granularity；
Creating unit, for the data information according to the loan application, creation obtains prediction model；
Predicting unit, the loan application amount for being treated each period in predicted time based on the prediction model are carried out
Prediction obtains the prediction result of the loan application amount of each period；
Assessment unit, for carrying out risk assessment to the loan application amount of each period according to the prediction result,
Obtain risk evaluation result.
Optionally, further includes:
Computing unit, if showing that the loan application amount of period to be predicted has exception for the risk evaluation result,
Calculate the loan application amount of the prediction period of the loan application amount for obtaining the prediction period and prediction model acquisition
Predicted quantity difference；
Determination unit, for determining fusing rank according to the difference；
Fuse unit, for according to the fusing rank, fuses to period to be predicted corresponding loan application
Processing is realized and is responded to the risk of the loan application.
Compared to the prior art, the present invention provides a kind of Risk Forecast Method and systems, are obtained and are borrowed based on time granularity
Then the data information of money application creates prediction model, such prediction model is created according to time series, Ke Yiyong
It predicts the loan application amount of each period, the result after prediction is subjected to risk assessment, Ke Yishi to practical loan application amount
Now the loan application amount of each period is assessed, then realizes real-time risk profile, and be based on time granularity, if
It concentrates or abnormal loan application can be identified, without manually being predicted, therefore, when solving artificial predicted
Low efficiency and the low problem of precision.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The embodiment of invention for those of ordinary skill in the art without creative efforts, can also basis
The attached drawing of offer obtains other attached drawings.
Fig. 1 is a kind of flow diagram of Risk Forecast Method provided in an embodiment of the present invention；
Fig. 2 is the flow diagram of another Risk Forecast Method provided in an embodiment of the present invention；
Fig. 3 is a kind of flow diagram of the creation method of prediction model provided in an embodiment of the present invention；
Fig. 4 is the process signal for the method that a kind of pair of loan application amount predicted value provided in an embodiment of the present invention is adjusted
Figure；
Fig. 5 is a kind of structural schematic diagram of Risk Forecast System provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Term " first " and " second " in description and claims of this specification and above-mentioned attached drawing etc. are for area
Not different objects, rather than for describing specific sequence.Furthermore term " includes " and " having " and their any deformations,
It is intended to cover and non-exclusive includes.Such as it contains the process, method of a series of steps or units, system, product or sets
It is standby not to be set in listed step or unit, but may include the step of not listing or unit.
A kind of Risk Forecast Method is provided in embodiments of the present invention, referring to Fig. 1, may comprise steps of:
S101, using the time as granularity, obtain the data information of loan application.
Wherein, the data information of loan application is primarily referred to as the loan application that loan system receives on some bank's line
Quantity information can also include the relevant information of loan application, for example, client-related information, interest rate information or policy are related
Information etc..It is to be acquired to be acquired at times to data information using the time as granularity, wherein the size of time granularity
It can be configured according to times of collection or acquisition interval, for example, being configured with times of collection, then will be acquired according to number
Time is divided, then can determine time granularity.Specifically, assume to need to acquire within 24 hours 8 times in one day, then at this time
Time granularity is 3 hours.
S102, the data information according to loan application, creation obtain prediction model；
According to the data information of acquisition, prediction model is established, which is used for in loan system on bank's line
Loan application is predicted with the presence or absence of risk.Prediction model in embodiments of the present invention can be arma modeling (Auto-
Regressive and Moving Average Model, autoregressive moving-average model), it is also possible to BP neural network mould
Type, or simultaneously create two models, choose wherein the best model of prediction effect as final target prediction model.Its
In, arma modeling is the important method of one kind of search time sequence, and BP neural network is that one kind inversely propagates calculation according to error
The multilayer feedforward neural network of method training.Certain embodiment of the present invention, which is only preferable over, obtains prediction mould using above two model
Type, but be not restricted to be predicted using other models, as long as the loan Shen of day part can be obtained based on prediction model
The prediction result that please be measure.
S103, treat the loan application amount of each period in predicted time based on prediction model and predicted, obtain with
The prediction result of the loan application amount of each period.
When treating the loan application amount of each period in predicted time based on prediction model and being predicted, for example, to certain
When it loan application amount is predicted, need to predict the loan application total amount on the same day first, then according to total amount and respectively
Period applications accounting, obtains the prediction result of day part client's applications.It can certainly be based on the Shen of history each period
It please measure, the applications predicted value for obtaining each period is directly predicted by prediction model.
And when being predicted within the time to be predicted, the calculated prediction knot of previous time granularity is not directlyed adopt
Fruit, but the practical loan application amount data of some periods in the predicted time are first obtained, then account for and work as according to these data
The ratio of the total amount of day adjusts the loan application quantity total amount of this day, finally treats subsequent each period in predicted time
The prediction result of loan application amount is adjusted, and can be adjusted in this way based on dynamic in real time, so that prediction result is more accurate,
And meet the actual conditions on the same day.
S104, according to prediction result, risk assessment is carried out to the loan application amount of each period, obtains risk assessment knot
Fruit.
Loan application quantity can be counted when loan system receives loan application on bank's line, be then based on this
The loan application amount prediction result of period carries out risk assessment to actual loan application quantity, obtains risk evaluation result.Example
Such as, the predicted value of the loan application quantity of some period is 50, and the loan application quantity actual value of the period is 150, then far
Far more than predicted value, the assessment result of high risk can be generated.It can be set when i.e. this carries out risk assessment according to predicted value certain
Certain risk class also can be set in floating range, the risk evaluation result of generation.
The present invention provides a kind of Risk Forecast Methods, the data information of loan application are obtained based on time granularity, then
Prediction model is created, such prediction model is created according to time series, can be used to predict the loan of each period
Result after prediction is carried out risk assessment to practical loan application amount, the loan Shen to each period may be implemented by applications
It please measure and be assessed, then realize real-time risk profile, and be based on time granularity, if concentration or abnormal loan Shen
It please can be identified, without manually being predicted, therefore, it is low solve manually low efficiency when being predicted and precision
Problem.
On the basis of Fig. 1 embodiment, the present invention also provides another Risk Forecast Methods, in this embodiment mainly
It is to be handled according to risk profile result, may include:
S201, using the time as granularity, obtain the data information of loan application；
S202, the data information according to loan application, creation obtain prediction model；
S203, treat the loan application amount of each period in predicted time based on prediction model and predicted, obtain with
The prediction result of the loan application amount of each period；
S204, according to prediction result, risk assessment is carried out to the loan application amount of each period, obtains risk assessment knot
Fruit；
S205, according to risk evaluation result, judge loan application amount with the presence or absence of abnormal, if not, execute S206, if
It is to execute S207；
S206, instruction loan application enter normal flow；
S207, the loan application amount and the loan application amount of the prediction period of prediction model acquisition for obtaining prediction period are calculated
Predicted quantity difference；
S208, according to difference, determine fusing rank；
S209, foundation fusing rank, treat the corresponding loan application of prediction period and carry out fusing processing, realize to loan Shen
Risk response please.
A fusing module can be set in the present embodiment will trigger the on-off module, to the loan when there are risk
Money application is intercepted, so as to make a response in time to events such as loan frauds.On the basis of Fig. 1 embodiment, this hair
Bright another embodiment further comprises a kind of creation method of prediction model, referring to Fig. 3, this method comprises:
S301, the data information according to loan application generate loan application amount time series；
S302, according to loan application amount time series, create the first prediction model；
S303, information extraction is carried out to the data information of loan application, obtains influence factor parameter information；
S304, it is based on influence factor parameter information, creates the second prediction model；
S305, the first prediction model of acquisition and the corresponding forecasting effective measure of the second prediction model are calculated separately；
S306, it is based on forecasting effective measure, target prediction model is determined in the first prediction model and the second prediction model.
Wherein, if the first prediction model is arma modeling, above-mentioned steps S302 can be specifically included:
According to loan application amount time series, the auto-correlation coefficient and partial correlation coefficient for obtaining sequence are calculated；
According to auto-correlation coefficient and partial correlation coefficient, the order of arma modeling is determined；
Using the arma modeling for having determined that order, the parameter of arma modeling is obtained；
Arma modeling is verified according to parameter, and is fitted based on verification result, several model of fit are obtained,
Target arma modeling is determined in model of fit.
If the second prediction model is BP neural network model, corresponding step S304 can be specifically included:
According to influence factor parameter information, the input sample collection of BP neural network model is determined；
According to practical loan application quantity, the output sample set of BP neural network model is established；
According to input sample collection and output sample set, establishes and obtain BP neural network model；
Parameter adjustment is carried out to BP neural network model, and carries out models fitting according to parameter adjusted, determines target
BP neural network model.
The prediction technique of the loan application amount proposed in the present invention is explained below, is treated based on prediction model
The loan application amount of each period in predicted time is predicted, the prediction result of the loan application amount of each period is obtained,
Include:
The loan application amount in predicted time is treated based on prediction model to be predicted, initial prediction is obtained；
Obtain the loan application amount actual numerical value of the first period collection in the time to be predicted；
Based on loan application amount actual numerical value, initial prediction is adjusted, when obtaining second in the time to be predicted
The prediction result of the loan application amount of each period of section collection, wherein the first period collection and the second period collection meet preset time
Corresponding relationship.
Specifically, referring to fig. 4, the side being adjusted for a kind of pair of loan application amount predicted value provided in an embodiment of the present invention
Method is based on loan application amount actual numerical value, is adjusted to initial prediction, obtains the second period collection in the time to be predicted
Each period loan application amount prediction result, comprising:
S401, the ratio that each period loan application amount accounts for loan application total amount is obtained；
S402, the loan application amount actual numerical value based on the first period collection and the corresponding ratio of the first period collection, meter
Calculate the calculated value for obtaining loan application total amount；
S403, according to the calculated value of the loan application total amount, treat the prediction of the loan application total amount in predicted time
Value is adjusted, and obtains target Prediction of Total value；
S404, the ratio that loan application total amount is accounted for according to the second period collection, when calculating second in the acquisition time to be predicted
The prediction result of the loan application amount of each period of section collection.
It should be noted that for meeting corresponding pass of preset time between above-mentioned the first period collection and the second period collection
System, the time for being primarily referred to as the first period collection can be earlier than the time that the second period concentrated, but for the first period collection and second
Between period collection time interval the present invention and with no restrictions.For example, it is desired to when the loan application amount to one day is predicted, the
One period, which integrated, to be the time range of morning 8:00 to 10:00, that is, needed to acquire the practical loan application amount of the time range
Data, and the second period collection can be any time range in the time from 10:00 to 24:00.Certain first period collection is simultaneously
It is not fixed and invariable, dynamic circulation selection can be carried out, the present invention is to this without repeating one by one.In the loan Shen to one day
It please measure when being predicted, not only need to predict the loan application total amount of this day, it is also desirable to the loan of this day each period
Money applications are predicted.And the prediction process is not a fixed process, but the mistake of a dynamic adjustment in real time
Journey, for example, need first to carry out prediction one prediction result of acquisition to t days on t-1 when predicting loan application amount in t days,
Then loan application total amount can be adjusted by obtaining the actual value of t days some periods, thus to the loan of remaining period
Money applications are adjusted.
Illustrate, it is assumed that t day client's total amount actual observed values be Yt, t=1,2,3 ..., T.WithIndicate t
The predicted value of day,For t days prediction errors；MtForecasting effective measure is indicated, to assess the prediction of distinct methods
Accuracy, observation are the true value of historical data.
Wherein, AtPrecision of prediction for prediction technique on t；[1, m] is forecast sample length of interval, includes m prediction
Point.
According to the characteristics of business client application, the prediction technique of use is as follows on line:
When selecting ARMA (p, q) model,
Yt=β1Yt-1+β2Yt-2+…+βpYt-p+εt+α1εt-1+α2εt-2+…+αqεt-q
Find out the auto-correlation coefficient (ACF) of the sequence of observations sample and the value of partial correlation coefficient (PACF).
ACF:
PACF: by
It can obtain:
ρ=P φ
Then
φ=P-1ρ
K=1,2 ... is substituted into above formula continuous solving, partial autocorrelation function can be acquired.
According to the property of sample autocorrelation coefficient and PARCOR coefficients, selecting order is when ARMA (p, q) model carries out
The β in model in value, that is, above-mentioned formula of unknown parameter is estimated in fitting1~βpAnd α1~αp
The validity of testing model is reappeared preference pattern if model of fit is not verified and is fitted.If quasi-
Molding type then needs to consider that other may establish multiple model of fit, from all model of fit by inspection by examining
Select optimal models.
Using the tendency in determining model of fit forecasting sequence future, and obtain the forecasting effective measure M of the model1。
When prediction model, which is based on BP neural network model, to be established, it is thus necessary to determine that influence factor is to establish model.
Resident's credit applications amount is influenced by credit applications wish, and credit applications wish and consumer behavior are closely bound up, by consumption of resident number
It is influenced according to, factors such as income data, true rate of interest.
If distinguishing town dweller and rural resident:
Consumption of resident data include cities and towns per capita consumption expenditure, rural area expenditure per capita；
Income data includes cities and towns per capita disposable income, per-capita net income in rural areas；
True rate of interest is that national 1 year deposit rate value subtracts 1 year deposit benefit of urban residents' consumption price index and the whole nation
Rate value subtracts rural residents income price index.
If not distinguishing town dweller and rural resident, consumption of resident data are that expenditure, income data are per capita
Per-capita gross domestic product, true rate of interest are that national 1 year fixed-term deposit rate value subtracts Consumer Prices index.
The collective effect for fully considering above-mentioned influence factor determines input sample collection and output sample set, establishes BP nerve
Network model, by adjusting model parameter, Optimal Fitting model, forecasting sequence by tendency, and obtain the prediction of the model
Availability M2。
According to M1And M2Size, select forecasting effective measure maximum and predicted as final prediction model.
Assuming that being divided into m sections of { P for daily 24 hours1,P2,…,Pm, withMinute is that unit counts the frequency, when each
The applications ratio of section is { P1,P2,…,Pm,It enablesThe t day obtained for optimum prediction model among the above
Total amount prediction result,For the applications prediction result of t days day parts, then:
Obtain at this time be t-1 days prediction t day at each moment applications prediction result.
On the day of t days, some period { P are obtained1,P2,…,PqApplications actual observed value { Yt1,Yt2,…Ytq, q
After=1,2 ... m-1, same day total amount ratio is accounted for according to these actual observed valuesT days total amounts of adjustment it is pre-
Measured valueAre as follows:
Then following sessions { P adjustedq+1,Pq+2,…,PmApplications prediction result adjustment are as follows:
The applications that can achieve the effect that the prediction day part of real-time dynamic self-adapting on line by the above process, are improved
The accuracy of prediction promotes the effect of early warning fusing system.
It should be noted that being adjusted to predicted value is because being divided into multiple periods daily, when some periods
After true value has, the former prediction result of following sessions can be carried out timely according to the difference of true value and predicted value
Adjustment improves accuracy.Fusing module has corresponding monitored data analysis, some of indexs and prediction application result and reality
The difference of applications is related.According to application record a demerit and the difference of practical applications beyond fusing module be arranged threshold value number come
Determine different fusing grades.
It is corresponding, the trigger mechanism that a kind of pair of applications predicted value is adjusted is additionally provided in embodiments of the present invention,
I.e. in the loan application amount actual numerical value according to the first period collection, after calculating the calculated value for obtaining loan application total amount, and it is different
Surely need to be adjusted every time, an adjustment threshold value can be set, by the difference of calculated value and predicted value and adjustment threshold value into
Row compares, and the Regulation mechanism in the embodiment of the present invention is triggered by comparing result, specifically, the trigger mechanism includes:
Calculate the calculated value of the loan application total amount and the predicted value of the loan application total amount in the time to be predicted
Between absolute difference；
Judge whether the absolute difference meets default adjustment threshold value, if it is, according to the loan application total amount
Calculated value, the predicted value for treating the loan application total amount in predicted time are adjusted, and obtain target Prediction of Total value.
Above-mentioned trigger mechanism is a kind of mode provided in an embodiment of the present invention, in addition it can be carried out based on other modes
The setting of triggering mode.For example, when monitoring that loan system occurs different on the line between other interconnected systems or branch, sub-line
Chang Shi can be after collecting practical loan application quantity in real time in order to ensure the normal work of loan system on the line of current row
Predicted value is adjusted, so that the predicted value and real-time collection value match to improve precision of prediction.Trigger mechanism i.e. at this time can be with
To be triggered according to the exception of predetermined system, the loan application total amount that can also be calculated is compared between total amount threshold value
Determine trigger mechanism, the present invention is to this without repeating one by one.
Risk Forecast Method provided in an embodiment of the present invention, being capable of the Shen of the prediction day part of dynamic self-adapting in real time on line
It please measure, improve the accuracy of prediction, improve the detection effect of early warning fusing system；By introducing adaptive prediction model,
So that bank can systematization real-time tracking client apply, draw, borrowing after etc. whole processes behavior, with Life cycle data product
It is tired, step up the accuracy for capturing client's unusual fluctuation behavior；It can reduce because wrong report causes business to be interrupted without reason, promoted simultaneously
While risk management validity, reduce to the bad experience of normal clients bring.
A kind of Risk Forecast System is additionally provided in embodiments of the present invention, referring to Fig. 5, comprising:
Acquiring unit 501, for obtaining the data information of loan application using the time as granularity；
Creating unit 502, for the data information according to loan application, creation obtains prediction model；
Predicting unit 503, the loan application amount for being treated each period in predicted time based on prediction model are carried out
Prediction, obtains the prediction result of the loan application amount of each period；
Assessment unit 504, for carrying out risk assessment to the loan application amount of each period, obtaining according to prediction result
Risk evaluation result.
The present invention provides a kind of Risk Forecast Systems, obtain the number of loan application based on time granularity in creating unit
It is believed that breath, then creates prediction model, such prediction model is created according to time series, can in predicting unit and
It is used to predict the loan application amount of each period in assessment unit, the result after prediction is subjected to risk to practical loan application amount
Assessment, may be implemented to assess the loan application amount of each period, then realizes real-time risk profile, and be to be based on
Time granularity, without manually being predicted, therefore, solves people if concentration or abnormal loan application can be identified
The low problem of low efficiency and precision when work is predicted.
On the basis of Fig. 5 embodiment optionally, further includes:
Computing unit calculates if showing that the loan application amount of period to be predicted has exception for risk evaluation result
The difference of the predicted quantity of the loan application amount for the prediction period that the loan application amount and prediction model for obtaining prediction period obtain；
Determination unit, for determining fusing rank according to difference；
Fuse unit, for treating the corresponding loan application of prediction period and carrying out fusing processing according to fusing rank, realizes
Risk response to loan application.
Optionally, the predicting unit includes:
Initial predicted subelement is carried out in advance for being treated the loan application amount in predicted time based on the prediction model
It surveys, obtains initial prediction；
Subelement is obtained, for obtaining the loan application amount actual numerical value of the first period collection in the time to be predicted；
Subelement is adjusted, for being based on the loan application amount actual numerical value, the initial prediction is adjusted, is obtained
Obtain the prediction result of the loan application amount of each period of the second period collection in the time to be predicted, wherein first period
The period summation of collection and the second period collection is the time to be predicted.
Wherein, the adjustment subelement is specifically used for:
Obtain the ratio that each period loan application amount accounts for loan application total amount；
Loan application amount actual numerical value and the corresponding ratio of the first period collection based on the first period collection are calculated and are obtained
The calculated value of loan application total amount；
According to the calculated value of the loan application total amount, the predicted value for treating the loan application total amount in predicted time is carried out
Adjustment obtains target Prediction of Total value；
The ratio of loan application total amount is accounted for according to the second period collection, is calculated and is obtained the second period collection in the time to be predicted
The prediction result of the loan application amount of each period.
Optionally, creating unit includes:
Sequence generates subelement, for the data information according to loan application, generates loan application amount time series；
First creation subelement, for creating the first prediction model according to loan application amount time series；
Subelement is extracted, information extraction is carried out for the data information to loan application, obtains influence factor parameter information；
Second creation subelement creates the second prediction model for being based on influence factor parameter information；
Computation subunit obtains the first prediction model and the corresponding prediction of the second prediction model effectively for calculating separately
Degree；
Model determines subelement, for being based on forecasting effective measure, determines in the first prediction model and the second prediction model
Target prediction model.
Optionally, the first creation subelement includes:
Coefficient computation subunit, for calculating the auto-correlation coefficient for obtaining sequence with according to loan application amount time series
And partial correlation coefficient；
Order determines subelement, for determining the order of arma modeling according to auto-correlation coefficient and partial correlation coefficient；
Parameter obtains subelement, for obtaining the parameter of arma modeling using the arma modeling for having determined that order；
First model determines subelement, carries out for being verified according to parameter to arma modeling, and based on verification result
Fitting, obtains several model of fit, and target arma modeling is determined in model of fit.
Optionally, the second creation subelement includes:
Input sample determines subelement, for determining the input of BP neural network model according to influence factor parameter information
Sample set；
Output sample determines subelement, for establishing the output of BP neural network model according to practical loan application quantity
Sample set；
Model foundation subelement, for establishing and obtaining BP neural network model according to input sample collection and output sample set；
Second model determines subelement, for carrying out parameter adjustment to BP neural network model, and according to ginseng adjusted
Number carries out models fitting, determines target BP neural network model.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.For device disclosed in embodiment
For, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is said referring to method part
It is bright.
The foregoing description of the disclosed embodiments enables those skilled in the art to implement or use the present invention.
Various modifications to these embodiments will be readily apparent to those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, of the invention
It is not intended to be limited to the embodiments shown herein, and is to fit to and the principles and novel features disclosed herein phase one
The widest scope of cause.
Claims (10)
1. a kind of Risk Forecast Method characterized by comprising
Using the time as granularity, the data information of loan application is obtained；
According to the data information of the loan application, creation obtains prediction model；
The loan application amount that each period in predicted time is treated based on the prediction model is predicted, is obtained described each
The prediction result of the loan application amount of period；
According to the prediction result, risk assessment is carried out to the loan application amount of each period, obtains risk evaluation result.
2. the method according to claim 1, wherein further include:
If the risk evaluation result shows that the loan application amount of period to be predicted has exception, when calculating the acquisition prediction
The difference of the predicted quantity of the loan application amount for the prediction period that the loan application amount of section and the prediction model obtain；
According to the difference, fusing rank is determined；
According to the fusing rank, fusing processing is carried out to period to be predicted corresponding loan application, is realized to the loan
The risk of money application responds.
3. the method according to claim 1, wherein described treated in predicted time based on the prediction model
The loan application amount of each period is predicted, the prediction result of the loan application amount of each period is obtained, comprising:
The loan application amount in predicted time is treated based on the prediction model to be predicted, initial prediction is obtained；
Obtain the loan application amount actual numerical value of the first period collection in the time to be predicted；
Based on the loan application amount actual numerical value, the initial prediction is adjusted, obtains in the time to be predicted
The prediction result of the loan application amount of each period of two period collection, wherein the first period collection and the second period collection
Meet preset time corresponding relationship.
4. according to the method described in claim 3, it is characterized in that, described be based on the loan application amount actual numerical value, to institute
It states initial prediction to be adjusted, obtains the prediction of the loan application amount of each period of the second period collection in the time to be predicted
As a result, comprising:
Obtain the ratio that each period loan application amount accounts for loan application total amount；
Loan application amount actual numerical value and the corresponding ratio of the first period collection based on the first period collection, calculating are provided a loan
The calculated value of total amount；
According to the calculated value of the loan application total amount, the predicted value for treating the loan application total amount in predicted time is adjusted
It is whole, obtain target Prediction of Total value；
The ratio of loan application total amount is accounted for according to the second period collection, is calculated and is obtained each of the second period collection in the time to be predicted
The prediction result of the loan application amount of period.
5. according to the method described in claim 4, it is characterized in that, being somebody's turn to do after calculating the calculated value for obtaining loan application total amount
Method further include:
It calculates between the calculated value of the loan application total amount and the predicted value of the loan application total amount in the time to be predicted
Absolute difference；
Judge whether the absolute difference meets default adjustment threshold value, if it is, according to the calculating of the loan application total amount
Value, the predicted value for treating the loan application total amount in predicted time are adjusted, and obtain target Prediction of Total value.
6. the method according to claim 1, wherein the data information according to the loan application, creation
Obtain prediction model, comprising:
According to the data information of the loan application, loan application amount time series is generated；
According to the loan application amount time series, the first prediction model is created；
Information extraction is carried out to the data information of the loan application, obtains influence factor parameter information；
Based on the influence factor parameter information, the second prediction model is created；
It calculates separately and obtains first prediction model and the corresponding forecasting effective measure of second prediction model；
Based on the forecasting effective measure, target prediction mould is determined in first prediction model and second prediction model
Type.
7. according to the method described in claim 6, it is characterized in that, when first prediction model is arma modeling, then institute
It states according to the loan application amount time series, creates the first prediction model, comprising:
According to the loan application amount time series, the auto-correlation coefficient and partial correlation coefficient for obtaining the sequence are calculated；
According to the auto-correlation coefficient and partial correlation coefficient, the order of arma modeling is determined；
Using the arma modeling for having determined that order, the parameter of the arma modeling is obtained；
The arma modeling is verified according to the parameter, and is fitted based on verification result, several fittings are obtained
Model determines target arma modeling in the model of fit.
8. according to the method described in claim 6, it is characterized in that, when second prediction model is BP neural network model
When, then it is described to be based on the influence factor parameter information, create the second prediction model, comprising:
According to the influence factor parameter information, the input sample collection of BP neural network model is determined；
According to practical loan application quantity, the output sample set of BP neural network model is established；
According to the input sample collection and the output sample set, establishes and obtain the BP neural network model；
Parameter adjustment is carried out to the BP neural network model, and carries out models fitting according to parameter adjusted, determines target
BP neural network model.
9. a kind of Risk Forecast System characterized by comprising
Acquiring unit, for obtaining the data information of loan application using the time as granularity；
Creating unit, for the data information according to the loan application, creation obtains prediction model；
Predicting unit, the loan application amount for treating each period in predicted time based on the prediction model carry out pre-
It surveys, obtains the prediction result of the loan application amount of each period；
Assessment unit, for carrying out risk assessment to the loan application amount of each period, obtaining according to the prediction result
Risk evaluation result.
10. system according to claim 6, which is characterized in that further include:
Computing unit calculates if showing that the loan application amount of period to be predicted has exception for the risk evaluation result
Obtain the loan application amount of the prediction period and the loan application amount of the prediction period that the prediction model obtains it is pre-
The difference of quantitation；
Determination unit, for determining fusing rank according to the difference；
Fuse unit, for according to the fusing rank, carries out fusing processing to period to be predicted corresponding loan application,
It realizes and the risk of the loan application is responded.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811085132.3A CN109166034A (en) | 2018-09-18 | 2018-09-18 | A kind of Risk Forecast Method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811085132.3A CN109166034A (en) | 2018-09-18 | 2018-09-18 | A kind of Risk Forecast Method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109166034A true CN109166034A (en) | 2019-01-08 |
Family
ID=64879617
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811085132.3A Pending CN109166034A (en) | 2018-09-18 | 2018-09-18 | A kind of Risk Forecast Method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109166034A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110135970A (en) * | 2019-04-15 | 2019-08-16 | 深圳壹账通智能科技有限公司 | Loan valuation method, apparatus, computer equipment and storage medium |
| CN110163470A (en) * | 2019-04-04 | 2019-08-23 | 阿里巴巴集团控股有限公司 | Case evaluating method and device |
| CN110347800A (en) * | 2019-07-15 | 2019-10-18 | 中国工商银行股份有限公司 | Text handling method and device and electronic equipment and readable storage medium storing program for executing |
| CN111815434A (en) * | 2020-07-10 | 2020-10-23 | 中国建设银行股份有限公司 | Credit protection method, device, equipment and storage medium |
| CN112819631A (en) * | 2021-02-10 | 2021-05-18 | 招联消费金融有限公司 | Service data processing method and device, computer equipment and storage medium |
| CN113095928A (en) * | 2021-04-08 | 2021-07-09 | 中国工商银行股份有限公司 | Real estate loan service risk assessment method and device |
| CN113919937A (en) * | 2021-09-22 | 2022-01-11 | 北京睿知图远科技有限公司 | KS monitoring system based on loan assessment wind control |
-
2018
- 2018-09-18 CN CN201811085132.3A patent/CN109166034A/en active Pending
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110163470A (en) * | 2019-04-04 | 2019-08-23 | 阿里巴巴集团控股有限公司 | Case evaluating method and device |
| CN110163470B (en) * | 2019-04-04 | 2023-05-30 | 创新先进技术有限公司 | Event evaluation method and device |
| CN110135970A (en) * | 2019-04-15 | 2019-08-16 | 深圳壹账通智能科技有限公司 | Loan valuation method, apparatus, computer equipment and storage medium |
| CN110347800A (en) * | 2019-07-15 | 2019-10-18 | 中国工商银行股份有限公司 | Text handling method and device and electronic equipment and readable storage medium storing program for executing |
| CN110347800B (en) * | 2019-07-15 | 2022-06-10 | 中国工商银行股份有限公司 | Text processing method and device, electronic equipment and readable storage medium |
| CN111815434A (en) * | 2020-07-10 | 2020-10-23 | 中国建设银行股份有限公司 | Credit protection method, device, equipment and storage medium |
| CN112819631A (en) * | 2021-02-10 | 2021-05-18 | 招联消费金融有限公司 | Service data processing method and device, computer equipment and storage medium |
| CN112819631B (en) * | 2021-02-10 | 2023-12-08 | 招联消费金融有限公司 | Service data processing method, device, computer equipment and storage medium |
| CN113095928A (en) * | 2021-04-08 | 2021-07-09 | 中国工商银行股份有限公司 | Real estate loan service risk assessment method and device |
| CN113919937A (en) * | 2021-09-22 | 2022-01-11 | 北京睿知图远科技有限公司 | KS monitoring system based on loan assessment wind control |
| CN113919937B (en) * | 2021-09-22 | 2023-06-23 | 北京睿知图远科技有限公司 | KS monitoring system based on loan assessment wind control |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109166034A (en) | A kind of Risk Forecast Method and system | |
| Keles et al. | Extended forecast methods for day-ahead electricity spot prices applying artificial neural networks | |
| CN108764863B (en) | Data transfer method, device, server and storage medium | |
| Babel et al. | A multivariate econometric approach for domestic water demand modeling: an application to Kathmandu, Nepal | |
| Khosravi et al. | Quantifying uncertainties of neural network-based electricity price forecasts | |
| CN110400021B (en) | Bank branch cash usage prediction method and device | |
| CN104881783A (en) | E-bank account fraudulent conduct and risk detecting method and system | |
| KR102009309B1 (en) | Management automation system for financial products and management automation method using the same | |
| CN107491970A (en) | Anti- cheating detection monitoring method and system and computing device in real time | |
| CN108550047A (en) | The prediction technique and device of trading volume | |
| Dao et al. | Analyzing the determinants of services trade flow between Vietnam and European Union: Gravity model approach | |
| Huang et al. | Contractor financial prequalification using simulation method based on cash flow model | |
| Matte et al. | Moving beyond the cost–loss ratio: economic assessment of streamflow forecasts for a risk-averse decision maker | |
| CN109583729B (en) | Data processing method and device for platform online model | |
| CN111408143A (en) | Game payment prediction method, model training method and device | |
| CN114943565A (en) | Electric power spot price prediction method and device based on intelligent algorithm | |
| Lu et al. | Analysis of early warning of RMB exchange rate fluctuation and value at risk measurement based on deep learning | |
| CN105005575A (en) | Quick developing interface method for enterprise intelligent prediction | |
| Rostamnia et al. | Investigating the effect of competitiveness power in estimating the average weighted price in electricity market | |
| CN113450158A (en) | Bank activity information pushing method and device | |
| CN113327160A (en) | Bank post-loan risk prediction method and system based on enterprise financial and capital big data | |
| CN116362136A (en) | Self-dispatching optimization method and system for independent energy storage system | |
| Chao et al. | Probabilistic approach to determining overhead-cum-markup rate in bid price | |
| Huang et al. | Multidimensional reputation evaluation model for crowdsourcing participants based on big data | |
| Talebi et al. | The Evaluation of the Capability of the Regression & Neural Network Models in Predicting Future Cash Flows |
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 |
Application publication date: 20190108 |
|
| RJ01 | Rejection of invention patent application after publication |