CN107093101A - Potential loan usage mining and risk score method based on POS pipelined datas - Google Patents
Potential loan usage mining and risk score method based on POS pipelined datas Download PDFInfo
- Publication number
- CN107093101A CN107093101A CN201710197360.9A CN201710197360A CN107093101A CN 107093101 A CN107093101 A CN 107093101A CN 201710197360 A CN201710197360 A CN 201710197360A CN 107093101 A CN107093101 A CN 107093101A
- Authority
- CN
- China
- Prior art keywords
- pos
- moon
- loan
- dealing money
- flowing water
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
Abstract
The invention discloses a kind of potential loan usage mining based on POS pipelined datas and risk score method, including：Obtain POS pipelined datas；These two aspects is started with terms of operation is expanded and in terms of capital turnover carries out potential loan usage mining to the POS pipelined datas of acquisition；It is determined that the statistical indicator for POS flowing water risk scores, and statistical indicator and Rating Model progress POS flowing water risk score of the POS pipelined datas obtained using setting according to determination.Start with terms of operation is expanded and in terms of capital turnover present invention incorporates POS pipelined datas and carry out potential loan usage mining, potential loan user can quickly and accurately be excavated, and POS pipelined datas embody trade company for fund and the demand of loan well, the success rate of conversion is higher；New POS flowing water risk score methods are proposed based on POS pipelined datas, it is more efficient.It the composite can be widely applied to Data Mining.
Description
Technical field
The present invention relates to Data Mining, especially a kind of potential loan usage mining based on POS pipelined datas with
Risk score method.
Background technology
At present in the electronic commerce times of market competition increasingly fierceness, more new clients are constantly expanded, from numerous numbers
Potential customers colony is excavated according to effective, and makes great efforts potential customers being converted into real client, enterprise just can obtain more multiple-effect
Benefit and market competition advantage.The purpose that potential customers excavate is exactly to formulate corresponding service strategy for enterprise to provide accurate ginseng
Examine foundation.
Existing loan user acquisition methods are most of to carry out ground popularization by marketing personnel, and not only efficiency is low, and
Successful conversion rate of providing a loan is not high.Although some potential user's method for digging have used computer technology, but due to being not bound with POS
Pipelined data, the accuracy rate of excavation is not often high.In addition, also lacking effective POS flowing water risk score side in the industry at present
Method, it would be highly desirable to further improve and improve.
The content of the invention
In order to solve the above technical problems, it is an object of the invention to：A kind of quick, accurate, loan successful conversion rate is provided
It is high and effective, potential loan usage mining and risk score method based on POS pipelined datas.
The technical solution used in the present invention is：
Potential loan usage mining and risk score method based on POS pipelined datas, comprise the following steps：
Obtain POS pipelined datas；
These two aspects is started with terms of operation is expanded and in terms of capital turnover carries out potential borrow to the POS pipelined datas of acquisition
Money usage mining；
It is determined that the statistical indicator for POS flowing water risk scores, and statistical indicator and the POS flowing water of acquisition according to determination
Data carry out POS flowing water risk scores using the Rating Model of setting.
Further, the POS pipelined datas are the POS pipelined datas of POS storages trade company, and the POS storages trade company refers to
POS has been installed, and had carried out at least one POS transaction, but has not yet carried out the user of target loan product application；It is described
POS storages trade company is also needed while meeting following 4 POS flowing water entry criteria：Industry access, at least six months transaction records,
At least six months dealing money are more than 5000 yuan in nearly 1 year and dealing money sequential growth rate is less than 500%.
Further, it is described from terms of expanding operation with capital turnover in terms of the POS flowing water numbers started with to acquisition of these two aspects
The step for according to potential loan usage mining is carried out, it includes：
The POS pipelined datas of acquisition are pre-processed, the pretreatment includes decompression, data cleansing, trading volume system
Meter and statistical result storage；
Operation aspect index analysis is enlarged according to pretreated POS pipelined datas；
Capital turnover aspect index analysis is carried out according to pretreated POS pipelined datas；
Result and the result of capital turnover aspect index analysis according to operation aspect index analysis is expanded judge that trade company is
It is no to have demand for loan.
It is further, described the step for be enlarged index analysis in terms of operation according to pretreated POS pipelined datas,
It is specially：
Pretreated POS pipelined datas are carried out using Kendall coefficient correlations moon dealing money trend, the moon to consume visitor
The judgement of amount trend and moon POS data trend, obtains corresponding moon dealing money Trend value, moon customer number Trend value
With moon POS data trend value.
Further, in terms of the progress capital turnover according to pretreated POS pipelined datas the step for index analysis,
It includes：
The moon equal dealing money average of credit card pen of all trade companies is calculated according to pretreated POS pipelined datas；
Moon credit card pen dealing money average is divided into according to the very site of every profession and trade POS data in all trade companies
Ten orderly intervals；
The trade company of application loan is classified according to the industry of trade company and ten intervals being divided into, so that application be borrowed
The trade company of money is respectively divided into ten intervals by respective industry；
The cut-point of the moon credit card equal dealing money average related to demand for loan is determined according to the result of classification.
Further, it is described according to the result for expanding operation aspect index analysis and the result of capital turnover aspect index analysis
The step for whether trade company has demand for loan judged, it is specially：
Judge whether trade company meets 5 any one had in demand for loan condition, if so, then judging that trade company has loan to need
Ask；Conversely, trade company is then judged without demand for loan, wherein, 5 have the demand for loan condition to be respectively：Month dealing money Trend value surpasses
Cross 0.5；The moon, customer volume trends value was more than 0.5；The moon, POS volume trends value was more than 0；Credit card pen dealing money is equal
It is worth 20% before ranking industry.
Further, the statistical indicator determined for POS flowing water risk scores, and statistical indicator according to determination and obtain
The step for POS pipelined datas taken carry out POS flowing water risk scores using the Rating Model of setting, it includes：
Statistical indicator for POS flowing water risk scores is filtered out using very site method and image method, it is described to be used for POS
The statistical indicator of flowing water risk score includes positive index, negative sense index and transaction moon number, and the positive index is handed over including the moon
Easy amount of money average, moon customer number average value, moon clearance stroke count accounting average and moon dealing money sequential growth rate Trend value,
The negative sense index includes moon credit card dealing money accounting average, the credit card odd-numbered day moon most dealing money multiplying power averages and the moon
Inquiry into balance stroke count multiplying power fluctuation；
It is determined that the score value computational methods and weight of the statistical indicator filtered out；
Entered according to the score value computational methods, weight and the POS of acquisition pipelined datas of determination using the Rating Model of setting
Row POS flowing water risk scores.
Further, the statistical indicator filtered out using very site method and image method for POS flowing water risk scores
The step for, it includes：
The very site of each statistical indicator every profession and trade is calculated by the historical data of all trade companies；
Each statistical indicator of POS storages trade company is mapped to ten orderly intervals using the very site of every profession and trade
In, wherein, POS storages trade company refers to install POS, and carries out at least one POS transaction, but not yet carries out target
The user of loan product application；
Choose cut-point and each statistical indicator of POS storages trade company is divided into orderly two sections or three sections, then
Using trade company history examination ＆ approval refusal or examine and each section of WOE values are calculated by information and IV values and draw corresponding image；
Optical sieving according to drawing goes out the statistical indicator for POS flowing water risk scores.
The step for score value computational methods and weight of the statistical indicator for determining to filter out, it includes：
Corresponding index score value computational methods are determined according to the industry ranking and classification of the statistical indicator filtered out, wherein,
Index score value computational methods are with 100 points for total score, and the index score value computational methods of the transaction month number are specially：If merchandising the moon
Number is less than 6 months, then its index score value is 0 point；If moon number of merchandising is more than 6 months and less than 24 months, often increased
One month, its index score increased by 5 points；If moon number of merchandising is more than 24 months, its index score value is 100 points；
The weight of corresponding index is determined according to the IV values and priori of the statistical indicator filtered out, wherein, the moon trade gold
The weight of volume average be 0.25, the moon credit card dealing money accounting average weight be 0.1, the credit card odd-numbered day moon most trade golds
The weight of volume multiplying power average be 0.05, the moon customer number average value weight be 0.15, the moon clear stroke count accounting average power
Weight is 0.15, and the weight of transaction moon number is 0.15, and the weight of moon inquiry into balance stroke count multiplying power fluctuation is 0.05, the moon trade gold
The weight of volume sequential growth rate Trend value is 0.1.
Further, it is described to use setting according to the score value computational methods, weight and the POS of acquisition pipelined datas of determination
The step for Rating Model carries out POS flowing water risk scores, it includes：
The corresponding all POS transaction data of trade company's numbering of an applicant are extracted from the POS pipelined datas of acquisition,
And all POS transaction data of extraction are merged into the POS pipelined datas of an applicant；
Judge applicant whether while 4 POS flowing water entry criteria are met, if so, then according to the score value calculating side of determination
Method, weight and the POS of acquisition pipelined datas calculate the score value of each statistical indicator using the Rating Model of setting, and by adding
Power obtains final POS flowing water risk scores；Conversely, final POS flowing water risk scores are then set to 0, wherein, 4 POS streams
Water entry criteria is respectively：Industry access, at least six months transaction records, at least six months dealing money surpass in nearly 1 year
Cross 5000 yuan and dealing money sequential growth rate is less than 500%.
The beneficial effects of the invention are as follows：Including obtain POS pipelined datas, from expand operation in terms of with capital turnover in terms of this
Two aspects, which are started with, to be carried out potential loan usage mining to the POS pipelined datas of acquisition and is referred to according to the POS flowing water of determination statistics
The step of mark and the POS pipelined datas obtained carry out POS flowing water risk scores using the Rating Model of setting, combines POS streams
Water number is started with according to these two aspects in terms of operation is expanded and in terms of capital turnover carries out potential loan usage mining, can be quick and accurate
Potential loan user really is excavated, and POS pipelined datas embody trade company for fund and the demand of loan well
Amount, the success rate of conversion is higher；New POS flowing water risk score methods are proposed based on POS pipelined datas, have been filled up in the industry
The blank of POS flowing water risk score methods, it is more efficient.
Brief description of the drawings
Fig. 1 is potential loan usage mining of the present invention based on POS pipelined datas and the overall flow of risk score method
Figure；
Fig. 2 is the POS pipelined data pretreatment process figures of the embodiment of the present invention one；
Fig. 3 is the cut-point schematic diagram of dealing money in January average of the embodiment of the present invention；
Fig. 4 is the cut-point schematic diagram of inquiry into balance in January stroke count multiplying power fluctuation of the embodiment of the present invention.
Embodiment
Reference picture 1, potential loan usage mining and risk score method based on POS pipelined datas, comprises the following steps：
Obtain POS pipelined datas；
These two aspects is started with terms of operation is expanded and in terms of capital turnover carries out potential borrow to the POS pipelined datas of acquisition
Money usage mining；
It is determined that the statistical indicator for POS flowing water risk scores, and statistical indicator and the POS flowing water of acquisition according to determination
Data carry out POS flowing water risk scores using the Rating Model of setting.
Wherein, POS flowing water risk score, is the credit risk for weighing applicant, and score value is higher, and risk is lower.
It is further used as preferred embodiment, the POS pipelined datas are the POS pipelined datas of POS storages trade company, institute
State POS storages trade company to refer to POS has been installed, and carried out at least one POS transaction, but not yet carry out target loan production
The user of product application；The POS storages trade company is also needed while meeting following 4 POS flowing water entry criteria：Industry access, at least
Have six months transaction records, in nearly 1 year at least six months dealing money more than 5000 yuan and dealing money sequential growth rate
Less than 500%.
Target loan product, refers to that enterprise treats the actual loan product recommended to user.
It is further used as preferred embodiment, described these two aspects is started with terms of operation is expanded and in terms of capital turnover
The step for potential loan usage mining is carried out to the POS pipelined datas of acquisition, it includes：
The POS pipelined datas of acquisition are pre-processed, the pretreatment includes decompression, data cleansing, trading volume system
Meter and statistical result storage；
Operation aspect index analysis is enlarged according to pretreated POS pipelined datas；
Capital turnover aspect index analysis is carried out according to pretreated POS pipelined datas；
Result and the result of capital turnover aspect index analysis according to operation aspect index analysis is expanded judge that trade company is
It is no to have demand for loan.
It is further used as preferred embodiment, it is described that operation side is enlarged according to pretreated POS pipelined datas
The step for face index analysis, it is specially：
Pretreated POS pipelined datas are carried out using Kendall coefficient correlations moon dealing money trend, the moon to consume visitor
The judgement of amount trend and moon POS data trend, obtains corresponding moon dealing money Trend value, moon customer number Trend value
With moon POS data trend value.
It is further used as preferred embodiment, it is described that capital turnover side is carried out according to pretreated POS pipelined datas
The step for face index analysis, it includes：
The moon equal dealing money average of credit card pen of all trade companies is calculated according to pretreated POS pipelined datas；
Moon credit card pen dealing money average is divided into according to the very site of every profession and trade POS data in all trade companies
Ten orderly intervals；
The trade company of application loan is classified according to the industry of trade company and ten intervals being divided into, so that application be borrowed
The trade company of money is respectively divided into ten intervals by respective industry；
The cut-point of the moon credit card equal dealing money average related to demand for loan is determined according to the result of classification.
It is further used as preferred embodiment, it is described according to the result for expanding operation aspect index analysis and capital turnover
The result of aspect index analysis judges the step for whether trade company has demand for loan, and it is specially：
Judge whether trade company meets 5 any one had in demand for loan condition, if so, then judging that trade company has loan to need
Ask；Conversely, trade company is then judged without demand for loan, wherein, 5 have the demand for loan condition to be respectively：Month dealing money Trend value surpasses
Cross 0.5；The moon, customer volume trends value was more than 0.5；The moon, POS volume trends value was more than 0；Credit card pen dealing money is equal
It is worth 20% before ranking industry.
It is further used as preferred embodiment, the statistical indicator determined for POS flowing water risk scores, and according to
The statistical indicator of determination and the POS pipelined datas obtained use this step of the Rating Model progress POS flowing water risk score of setting
Suddenly, it includes：
Statistical indicator for POS flowing water risk scores is filtered out using very site method and image method, it is described to be used for POS
The statistical indicator of flowing water risk score includes positive index, negative sense index and transaction moon number, and the positive index is handed over including the moon
Easy amount of money average, moon customer number average value, moon clearance stroke count accounting average and moon dealing money sequential growth rate Trend value,
The negative sense index includes moon credit card dealing money accounting average, the credit card odd-numbered day moon most dealing money multiplying power averages and the moon
Inquiry into balance stroke count multiplying power fluctuation；
It is determined that the score value computational methods and weight of the statistical indicator filtered out；
Entered according to the score value computational methods, weight and the POS of acquisition pipelined datas of determination using the Rating Model of setting
Row POS flowing water risk scores.
It is further used as preferred embodiment, it is described to be filtered out using very site method and image method for POS flowing water
The step for statistical indicator of risk score, it includes：
The very site of each statistical indicator every profession and trade is calculated by the historical data of all trade companies；
Each statistical indicator of POS storages trade company is mapped to ten orderly intervals using the very site of every profession and trade
In, wherein, POS storages trade company refers to install POS, and carries out at least one POS transaction, but not yet carries out target
The user of loan product application；
Choose cut-point and each statistical indicator of POS storages trade company is divided into orderly two sections or three sections, then
Using trade company history examination ＆ approval refusal or examine and each section of WOE values are calculated by information and IV values and draw corresponding image；
Optical sieving according to drawing goes out the statistical indicator for POS flowing water risk scores.
The step for score value computational methods and weight of the statistical indicator for determining to filter out, it includes：
Corresponding index score value computational methods are determined according to the industry ranking and classification of the statistical indicator filtered out, wherein,
Index score value computational methods are with 100 points for total score, and the index score value computational methods of the transaction month number are specially：If merchandising the moon
Number is less than 6 months, then its index score value is 0 point；If moon number of merchandising is more than 6 months and less than 24 months, often increased
One month, its index score increased by 5 points；If moon number of merchandising is more than 24 months, its index score value is 100 points；
The weight of corresponding index is determined according to the IV values and priori of the statistical indicator filtered out, wherein, the moon trade gold
The weight of volume average be 0.25, the moon credit card dealing money accounting average weight be 0.1, the credit card odd-numbered day moon most trade golds
The weight of volume multiplying power average be 0.05, the moon customer number average value weight be 0.15, the moon clear stroke count accounting average power
Weight is 0.15, and the weight of transaction moon number is 0.15, and the weight of moon inquiry into balance stroke count multiplying power fluctuation is 0.05, the moon trade gold
The weight of volume sequential growth rate Trend value is 0.1.
It is further used as preferred embodiment, the score value computational methods, weight and the POS of acquisition according to determination
The step for pipelined data carries out POS flowing water risk scores using the Rating Model of setting, it includes：
The corresponding all POS transaction data of trade company's numbering of an applicant are extracted from the POS pipelined datas of acquisition,
And all POS transaction data of extraction are merged into the POS pipelined datas of an applicant；
Judge applicant whether while 4 POS flowing water entry criteria are met, if so, then according to the score value calculating side of determination
Method, weight and the POS of acquisition pipelined datas calculate the score value of each statistical indicator using the Rating Model of setting, and by adding
Power obtains final POS flowing water risk scores；Conversely, final POS flowing water risk scores are then set to 0, wherein, 4 POS streams
Water entry criteria is respectively：Industry access, at least six months transaction records, at least six months dealing money surpass in nearly 1 year
Cross 5000 yuan and dealing money sequential growth rate is less than 500%.
The present invention is further explained and illustrated with reference to Figure of description and specific embodiment.
Embodiment one
Low for existing potential loan usage mining method efficiency, loan successful conversion rate is not high, and the accuracy rate of excavation is low
And the problem of lack effective POS flowing water risk score method, the present invention proposes a kind of new based on POS flowing water numbers
According to potential loan usage mining and risk score method, can quickly and accurately excavate potential loan user, and POS
Flowing water can embody trade company well for fund and the demand of loan, and the success rate of conversion is higher.
Below from data source, potential trade company excavate and POS flowing water risk score statistical indicators with calculate this 3 aspect pair
The realization principle of the present invention is described in detail.
(1) data source
The data that the present invention is excavated are both from POS storages trade company, and POS storages trade company refers to install POS, and enters
At least one POS of going is merchandised, but not yet carries out the user of target loan product application.And these POS storages trade companies need
Following 4 POS flowing water entry criteria are met simultaneously：
The transaction records of a, at least six months；
B, industry access；
C, in nearly 1 year at least six months dealing money more than 5000 yuan；
D, dealing money sequential growth rate are less than 500%.
(2) potential trade company excavates
The present invention excavates potential trade company in terms of following two：
(1) in terms of expanding operation：Trade company is to be badly in need of loan when needing and expanding and manage, therefore if trade company is in transaction
There is growth trend in the amount of money, customer quantity and POS quantity, it is of the invention then think that trade company has and expand the loan managed and need
Ask.
Wherein, the present invention need to use Kendall tau orders when being analyzed in terms of being enlarged operation according to POS pipelined datas
Coefficient correlation is traded the judgement of amount of money trend, moon customer number trend, moon POS data trend.
Kendall tau values are between -1 and+1.The value of Kendall coefficient correlations is higher, and correlation degree is stronger.One
As for, when Kendall coefficients are 0.9 or more, it is very strong to be considered as positive incidence degree, that is, positive tendency is non-
Chang Qiang.Higher or significant Kendall coefficients mean to employ basically identical standard when inspector assesses sample.When
When Kendall coefficients are less than -0.9, illustrate that reverse correlation program is very strong, that is, reverse tendency is very strong.Work as Kendall
When coefficient is close to 0, illustrate association is not present, that is to say, that no trend can be sayed.
Kendall tau coefficients are commonly used for following occasion：
(1) with the linear relationship between non-parametric test method measurement ordinal data；
(2) it is consistent to number and non-uniform to number according to calculating using variable rank number.
1) it is when two variables have stronger positive correlation, then consistent larger to number, it is non-uniform smaller to number；
2) it is when two variables have stronger negative correlativing relation, then consistent smaller to number, it is non-uniform larger to number
3) it is when two correlation of variables are weaker, then consistent to number and non-uniform roughly equal to number.
Kendall tau coefficients τ computational methods are：
Wherein, U is to be consistent to number, and V is non-uniform to number, and n is the quantity of sample,For total logarithm.
Under small sample, Kendall coefficient correlations obey Kendall distributions；Under large sample, Kendall coefficient correlations
Test statistics be Z statistics, Z statistics are defined as：
Wherein, Z statistics approximately obey standardized normal distribution.
Calculated by Kendall tau rank correlation coefficients the moon dealing money Trend value, moon customer volume trends
Value and moon POS volume trends value are all between -1 to 1.By statistics, it is found that about 10% trade company month dealing money becomes
And comparatively gesture value, moon customer volume trends value are more than 0.5, Kendall rank correlation coefficients more than 0.5
Compare high, therefore the of that month dealing money Trend value of potential trade company's mining model setting or moon customer quantity of the present invention
When Trend value is more than 0.5, the demand for loan in terms of there is expansion operation in trade company is considered as.And for moon POS volume trends value
For, 0.5 boundary is too high, because most trade companies just only have one to two POSs, the quantity increase of POS is less
Readily, thus the present invention potential trade company's mining model set moon POS volume trends value more than 0 when, be considered as trade company
There is the demand for loan in terms of expansion operation.
(2) in terms of capital turnover：Trade company may need loan to carry out capital turnover due to short of money, for those individuals
For trade company, bank loan is relatively difficult, and trade company is likely to use credit card arbitrage, and if letter in trade company's POS
Consume averagely too many beyond industry with card pen, then more likely think the existing behavior of the creditable cutting ferrule of trade company, reflect trade company
There is the demand for loan of capital turnover.The present invention passes through the result of calculation to POS storages trade company historical data and application loan trade company
Performance contrasted, it is determined that the cut-point of the statistical indicator related to demand for loan.
Analysis indexes selection in terms of capital turnover of the present invention is moon credit card pen dealing money average.Month credit card pen
Equal dealing money average needs in industry compare in portion, therefore the present invention is first according to ten of every profession and trade POS data in POS storages trade company
Quantile by moon credit card pen dealing money average be divided into orderly ten intervals (by arriving big order from childhood, from 1 to
10), the trade company of loan then will be applied according in respective trade division to this ten intervals.It can be drawn by observation, the 9th
Substantially exceed 10% with 10 the two interval trade company's quantity accountings, illustrate most latter two interval store credit cards arbitrage
Possibility is larger.
(3) finally, potential trade company's mining model of the invention determines following rule, as long as meeting any one therein,
Being considered as trade company has demand for loan：
The moon, dealing money Trend value was more than 0.5；
The moon, customer volume trends value was more than 0.5；
The moon, POS volume trends value was more than 0；
Preceding the 20% of credit card pen dealing money average ranking industry.
When carrying out potential trade company's excavation, in addition it is also necessary to first POS pipelined datas are pre-processed, the flow of pretreatment is as schemed
Shown in 2.
(3) POS flowing water risk score statistical indicator and calculation specifications
(1) POS flowing water risk score statistical indicator is selected
By taking all trade companies in Guangdong Province as an example, the selection of POS flowing water risk scores statistical indicator specifically includes procedure below：
1) the very site of each statistical indicator every profession and trade is calculated using the historical data of all trade companies of Guangdong Province；
2) to each statistical indicator, POS storages trade company is mapped to 1 to 10 this ten using the very site of every profession and trade
In interval (totally 10 sections, arrange from low to high)；
3) to each statistical indicator, choose cut-point and be divided into orderly two sections or three sections, then deposited using POS
The history examination ＆ approval refusal of amount trade company is examined and calculates each section of WOE values by information and IV values and draw corresponding image；
4) statistical indicator for POS flowing water risk scores is gone out according to the optical sieving drawn.The present invention filters out information
It is worth larger and significant statistical indicator and is used as Score index.
Wherein, IV full name is Information Value, and Chinese means information value, or information content.WOE's
Full name is Weight of Evidence, i.e. evidence weight.WOE is a kind of coding form to original argument, and it is actual just
It is that independent variable takes a kind of influence on ratio of breaking a contract when some value.The present invention has continued to use existing when calculating WOE values and IV values
Some WOE values and IV value calculation formula.
(2) 8 filtered out are used for the statistical indicator introduction of POS flowing water risk scores
The present invention filters out eight use altogether according to index selection course and statistical distribution and priori in (three) (1)
In the statistical indicator of POS flowing water risk scores, this 8 indexs are specially：
1) moon dealing money average
The definition of month dealing money average：Each moon dealing money summation of all trade companies of applicant's correspondence is calculated, so
After be averaged.
For example：If the moon, dealing money was as shown in table 1 below, the moon dealing money average=(100+200+300)/3=200
Table 1
| Month | January | 2 months | March |
| Month dealing money | 100 | 200 | 300 |
2) moon credit card dealing money accounting average
The definition of month credit card dealing money accounting：Month credit card dealing money accounting=moon credit card dealing money/moon
Dealing money.
For example, the moon dealing money of some trade company is 200, the moon credit card dealing money be 100, then, its month credit card
Dealing money accounting=100/200=0.5.
Month credit card dealing money accounting average, i.e., be averaged to moon credit card dealing money accounting.
For example, the moon credit card dealing money accounting it is as shown in table 2 below, then, the moon credit card dealing money accounting average=
(0.2+0.1+0.3)/3=0.2.
Table 2
3) the credit card odd-numbered day moon most dealing money multiplying power averages
The definition of month credit card odd-numbered day most dealing money multiplying powers：Month credit card odd-numbered day most dealing money multiplying power=menses
/ (moon credit card dealing money/number of days of merchandising the moon with card odd-numbered day most dealing money).
For example, certain trade company merchandises 30 days April altogether, and have and have within 3 days a moon credit card transaction record, credit card trade stroke count is detailed
It is as shown in table 3 below：
Table 3
| X days certain month | Y days certain month | Z days certain month | |
| The credit card trade amount of money | 10 | 15 | 5 |
So, then have：
Moon transaction number of days=30；
Month credit card dealing money=10+15+5=30；
Month credit card odd-numbered day most dealing money=15；
Month credit card odd-numbered day most dealing money multiplying power=15/ (30/30)=15.
And the credit card odd-numbered day moon most dealing money multiplying power averages, i.e., each credit card odd-numbered day most dealing money multiplying powers month in and month out
It is averaged.
4) moon inquiry into balance stroke count multiplying power fluctuation
The definition of month inquiry into balance stroke count multiplying power：Month inquiry into balance stroke count multiplying power=moon inquiry into balance stroke count/moon transaction pen
Number.
For example, moon transaction stroke count is 100, moon inquiry into balance stroke count is 60, then moon inquiry into balance stroke count multiplying power=60/100
=0.6.
Month inquiry into balance stroke count multiplying power fluctuation, the i.e. standard deviation of inquiry into balance every month stroke count multiplying power, for weighing it
Fluctuate size.
For example, trade company A moon inquiry into balance stroke count multiplying power is as shown in table 4 below, because inquiry into balance every month stroke count multiplying power
It is all identical, show it without fluctuation, so moon inquiry into balance stroke count multiplying power fluctuation=0.
Table 4
| Month | January | 2 months | March |
| Month inquiry into balance stroke count multiplying power | 0.6 | 0.6 | 0.6 |
5) moon dealing money sequential growth rate Trend value
The definition of month dealing money sequential growth rate：Month dealing money sequential growth rate=(second month dealing money-the
One month dealing money)/first month dealing money.
For example, dealing money in March is 300, April, dealing money was 600, the moon dealing money sequential growth rate=(600-
300)/300=1.
Month dealing money sequential growth rate Trend value is related to the time for weighing moon dealing money sequential growth rate
Property, it uses kendall rank correlation coefficients to calculate.If over the passage of time, growth rate is increasing, the moon dealing money
Sequential growth rate Trend value=1；If over the passage of time, growth rate is less and less, the moon dealing money sequential growth rate become
Gesture value=- 1.Month dealing money sequential growth rate Trend value is always between -1 to 1.
6) moon customer number average value
The definition of month customer number average value：The average difference every month customer quantity of trade company.
7) moon clearance stroke count accounting average
The moon clears the definition of stroke count accounting：Clear the moon and clear stroke count/moon total stroke count by stroke count accounting=moon.
The moon clears stroke count accounting average, i.e., each stroke count accounting of clearance month in and month out is averaged.
8) transaction moon number
The definition of transaction moon number：Applicant has the moon number of transaction record from transaction is started untill the application time.Such as
Fruit has do not have transaction record some months halfway, then these months do not count.
(3) it is used for the statistical indicator classification of POS flowing water risk scores
Statistical indicator for POS flowing water risk scores is divided into positive index, negative sense index and transaction month by the present invention
This 3 class of number.
Wherein, positive index：Index value is bigger, and examination ＆ approval more tendency passes through；Index value is smaller, and examination ＆ approval more tendency is refused
Absolutely.Month dealing money average, moon customer number average value, moon clearance stroke count accounting average and moon dealing money sequential growth rate
Trend value belongs to positive index.
Negative sense index：Index value is bigger, examination ＆ approval more tendency refusal；Index value is smaller, and examination ＆ approval more tendency passes through.Menses
Fluctuated with card dealing money accounting average, the credit card odd-numbered day moon most dealing money multiplying power averages and moon inquiry into balance stroke count multiplying power
Property belongs to negative sense index.
For positive index, by taking moon dealing money average as an example, 70% quantile is chosen as cut-point, Fig. 3 is can obtain.
In Fig. 3, first paragraph 1 be ranking after every profession and trade 70% trade company, second segment 2 be ranking before every profession and trade 30% trade company, indulge
Axle represents accounting of the Ge Duan trade companies in correspondence classification trade company, as the buttress shaft body surface of first paragraph 1 shows ranking after every profession and trade 70%
The trade company being rejected in trade company accounts for the overall ratio of refusal trade company.From figure 3, it can be seen that moon dealing money average and approval results
There is obvious positive relationship.
For negative sense index, by taking moon inquiry into balance stroke count multiplying power fluctuation as an example, 60% quantile is chosen as cut-point,
It can obtain Fig. 4.In Fig. 4, first paragraph 1 be ranking after every profession and trade 60% trade company, second segment 2 be ranking 40% before every profession and trade
Trade company, the longitudinal axis represents accounting of the Ge Duan trade companies in correspondence classification trade company, and such as buttress shaft body surface of first paragraph 1 shows ranking in every profession and trade
The trade company being rejected afterwards in 60% trade company accounts for the overall ratio of refusal trade company.From fig. 4, it can be seen that moon inquiry into balance stroke count times
Rate fluctuation has obvious negative sense relation with approval results.
(4) the score calculation explanation of Rating Model statistical indicator
The score value of Rating Model statistical indicator is calculated：Each index is that total score is carried out continuously with 100 points according to industry ranking
Marking.For example, before industry 1% give 99 points or 1 point, before industry 12% to 88 points or 12 points, it is necessary to root
Determine to take positive score value (99 points, 88 points) or negative sense score value (1 point, 12 points) according to the classification of specific targets.
The weight of statistical indicator is determined：Preliminary weight is determined according to the weighting of the IV values of index, then entered according to priori
Row adjustment.
The statistical indicator and its weight for POS flowing water risk scores that the present embodiment is finally determined are as shown in table 6 below：
Table 6
(5) POS flowing water risk score
POS flowing water risk score is for single trade company numbers in theory, actually general loan application people
Multiple trade company's numberings can be provided, need to be unified into a trade company when POS flowing water risk scores are carried out.Therefore, it is of the invention
Specific POS flowing water risk score calculating process is as follows：
1) (one or more) are numbered for the trade company that loan application people provides, extracts trade company and number corresponding all POS
Transaction data, and they are merged into the data processing of a loan application people；
2) trade company's POS Flow Records of industry not access are rejected (if after rejecting being empty, its in (one) data source
Its 3 POS flowing water entry criteria is unsatisfactory for automatically)；
3) judge whether other 3 POS flowing water entry criteria meet simultaneously in (one) data source, as long as there is an access
Condition is unsatisfactory for, then final POS flowing water risk score is 0；
4) for meeting the loan application people of entry criteria, point of each index is calculated according to Rating Model set in advance
It is worth and weights and obtains final POS flowing water risk scores.
Above is the preferable implementation to the present invention is illustrated, but the present invention is not limited to the embodiment, ripe
A variety of equivalent variations or replacement can also be made on the premise of without prejudice to spirit of the invention by knowing those skilled in the art, this
Equivalent deformation or replacement are all contained in the application claim limited range a bit.
Claims (10)
1. potential loan usage mining and risk score method based on POS pipelined datas, it is characterised in that：Including following step
Suddenly：
Obtain POS pipelined datas；
These two aspects, which is started with, in terms of operation is expanded and in terms of capital turnover carries out potential loan use to the POS pipelined datas of acquisition
Excavate at family；
It is determined that the statistical indicator for POS flowing water risk scores, and statistical indicator and the POS pipelined datas of acquisition according to determination
POS flowing water risk scores are carried out using the Rating Model of setting.
2. the potential loan usage mining according to claim 1 based on POS pipelined datas and risk score method, it is special
Levy and be：The POS pipelined datas are the POS pipelined datas of POS storages trade company, and the POS storages trade company refers to install POS
Machine, and at least one POS transaction was carried out, but not yet carry out the user of target loan product application；The POS storages business
Family is also needed while meeting following 4 POS flowing water entry criteria：Industry access, at least six months transaction records, in nearly 1 year extremely
Rare six months dealing money are more than 5000 yuan and dealing money sequential growth rate is less than 500%.
3. the potential loan usage mining according to claim 1 based on POS pipelined datas and risk score method, it is special
Levy and be：The these two aspects in terms of operation is expanded and in terms of capital turnover is started with is dived to the POS pipelined datas of acquisition
The step for usage mining is provided a loan, it includes：
The POS pipelined datas of acquisition are pre-processed, it is described pretreatment include decompression, data cleansing, trading volume count and
Statistical result is put in storage；
Operation aspect index analysis is enlarged according to pretreated POS pipelined datas；
Capital turnover aspect index analysis is carried out according to pretreated POS pipelined datas；
Result and the result of capital turnover aspect index analysis according to operation aspect index analysis is expanded judge whether trade company has
Demand for loan.
4. the potential loan usage mining according to claim 3 based on POS pipelined datas and risk score method, it is special
Levy and be：Described the step for be enlarged index analysis in terms of operation according to pretreated POS pipelined datas, its is specific
For：
Moon dealing money trend, moon customer number are carried out to pretreated POS pipelined datas using Kendall coefficient correlations
The judgement of trend and moon POS data trend, obtains corresponding moon dealing money Trend value, moon customer number Trend value and the moon
POS data trend value.
5. the potential loan usage mining according to claim 4 based on POS pipelined datas and risk score method, it is special
Levy and be：In terms of the progress capital turnover according to pretreated POS pipelined datas the step for index analysis, it includes：
The moon equal dealing money average of credit card pen of all trade companies is calculated according to pretreated POS pipelined datas；
Moon credit card pen dealing money average is divided into order according to the very site of every profession and trade POS data in all trade companies
Ten intervals；
The trade company of application loan is classified according to the industry of trade company and ten intervals being divided into, so that loan will be applied for
Trade company is respectively divided into ten intervals by respective industry；
The cut-point of the moon credit card equal dealing money average related to demand for loan is determined according to the result of classification.
6. the potential loan usage mining according to claim 5 based on POS pipelined datas and risk score method, it is special
Levy and be：The result according to the result for expanding operation aspect index analysis and capital turnover aspect index analysis judges trade company
The step for whether having demand for loan, it is specially：
Judge whether trade company meets 5 any one had in demand for loan condition, if so, then judging that there is demand for loan in trade company；
Conversely, trade company is then judged without demand for loan, wherein, 5 have the demand for loan condition to be respectively：Month dealing money Trend value exceedes
0.5；The moon, customer volume trends value was more than 0.5；The moon, POS volume trends value was more than 0；Credit card pen dealing money average
20% before ranking industry.
7. potential loan usage mining based on POS pipelined datas and risk score according to claim any one of 1-6
Method, it is characterised in that：The statistical indicator determined for POS flowing water risk scores, and statistical indicator according to determination and
The step for POS pipelined datas of acquisition carry out POS flowing water risk scores using the Rating Model of setting, it includes：
Statistical indicator for POS flowing water risk scores is filtered out using very site method and image method, it is described to be used for POS flowing water
The statistical indicator of risk score includes positive index, negative sense index and transaction moon number, and the positive index includes moon trade gold
Volume average, moon customer number average value, moon clearance stroke count accounting average and moon dealing money sequential growth rate Trend value, it is described
Negative sense index includes moon credit card dealing money accounting average, the credit card odd-numbered day moon most dealing money multiplying power averages and moon remaining sum
Inquire about stroke count multiplying power fluctuation；
It is determined that the score value computational methods and weight of the statistical indicator filtered out；
POS is carried out using the Rating Model of setting according to the score value computational methods, weight and the POS of acquisition pipelined datas of determination
Flowing water risk score.
8. the potential loan usage mining according to claim 7 based on POS pipelined datas and risk score method, it is special
Levy and be：It is described to use the step for very site method and image method filter out the statistical indicator for POS flowing water risk scores,
It includes：
The very site of each statistical indicator every profession and trade is calculated by the historical data of all trade companies；
Each statistical indicator of POS storages trade company is mapped in ten orderly intervals using the very site of every profession and trade,
Wherein, POS storages trade company refers to install POS, and carries out at least one POS transaction, but not yet carries out target loan
The user of money request for product；
Choose cut-point and each statistical indicator of POS storages trade company is divided into orderly two sections or three sections, then utilize
The history examination ＆ approval refusal of trade company is examined and calculates each section of WOE values by information and IV values and draw corresponding image；
Optical sieving according to drawing goes out the statistical indicator for POS flowing water risk scores.
9. the potential loan usage mining according to claim 8 based on POS pipelined datas and risk score method, it is special
Levy and be：The step for score value computational methods and weight of the statistical indicator for determining to filter out, it includes：
Corresponding index score value computational methods are determined according to the industry ranking and classification of the statistical indicator filtered out, wherein, index
Score value computational methods are with 100 points for total score, and the index score value computational methods of the transaction month number are specially：If merchandising moon number
For less than 6 months, then its index score value was 0 point；If moon number of merchandising was more than 6 months and less than 24 months, often increase by one
Month, its index score increases by 5 points；If moon number of merchandising is more than 24 months, its index score value is 100 points；
The weight of corresponding index is determined according to the IV values and priori of the statistical indicator filtered out, wherein, the moon dealing money it is equal
The weight of value is 0.25, the moon weight of credit card dealing money accounting average be 0.1, the credit card odd-numbered day moon most dealing money times
The weight of rate average be 0.05, the moon customer number average value weight be 0.15, the moon clearance stroke count accounting average weight be
0.15, the weight of transaction moon number is 0.15, and the weight of moon inquiry into balance stroke count multiplying power fluctuation is 0.05, the moon dealing money ring
Weight than growth rate Trend value is 0.1.
10. potential loan usage mining based on POS pipelined datas according to claim 8 or claim 9 and risk score method,
It is characterized in that：It is described that the scoring set is used according to the score value computational methods, weight and the POS of acquisition pipelined datas of determination
The step for model carries out POS flowing water risk scores, it includes：
The corresponding all POS transaction data of trade company's numbering of an applicant are extracted from the POS pipelined datas of acquisition, and will
All POS transaction data extracted merge into the POS pipelined datas of an applicant；
Judge applicant whether while 4 POS flowing water entry criteria are met, if so, then according to the score value computational methods of determination, power
Weight and the POS pipelined datas obtained calculate the score value of each statistical indicator using the Rating Model of setting, and by weighting
To final POS flowing water risk scores；Conversely, final POS flowing water risk scores are then set to 0, wherein, 4 POS flowing water are accurate
Entering condition is respectively：Industry access, at least six months transaction records, at least six months dealing money exceed in nearly 1 year
5000 yuan and dealing money sequential growth rate are less than 500%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710197360.9A CN107093101A (en) | 2017-03-29 | 2017-03-29 | Potential loan usage mining and risk score method based on POS pipelined datas |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710197360.9A CN107093101A (en) | 2017-03-29 | 2017-03-29 | Potential loan usage mining and risk score method based on POS pipelined datas |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107093101A true CN107093101A (en) | 2017-08-25 |
Family
ID=59646292
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710197360.9A Pending CN107093101A (en) | 2017-03-29 | 2017-03-29 | Potential loan usage mining and risk score method based on POS pipelined datas |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107093101A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109636577A (en) * | 2018-10-25 | 2019-04-16 | 深圳壹账通智能科技有限公司 | IP address analysis method, device, equipment and computer readable storage medium |
| CN109976934A (en) * | 2019-02-26 | 2019-07-05 | 上海拍拍贷金融信息服务有限公司 | Equipment uses abnormality determination method, device and computer storage medium |
| CN110197316A (en) * | 2018-06-14 | 2019-09-03 | 腾讯科技（深圳）有限公司 | Processing method, device, computer-readable medium and the electronic equipment of operation data |
| CN110414914A (en) * | 2019-06-17 | 2019-11-05 | 深圳壹账通智能科技有限公司 | Business datum monitoring method and device |
| CN113313572A (en) * | 2021-05-28 | 2021-08-27 | 上海浦东发展银行股份有限公司 | Model identification method based on accumulation fund point-credit customer |
| CN114693188A (en) * | 2022-05-31 | 2022-07-01 | 四川骏逸富顿科技有限公司 | Risk supervision method, system and equipment for drug retail industry |
| CN117708183A (en) * | 2023-11-08 | 2024-03-15 | 广州西米科技有限公司 | Potential user mining method and system based on user consumption habit |
-
2017
- 2017-03-29 CN CN201710197360.9A patent/CN107093101A/en active Pending
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110197316A (en) * | 2018-06-14 | 2019-09-03 | 腾讯科技（深圳）有限公司 | Processing method, device, computer-readable medium and the electronic equipment of operation data |
| CN110197316B (en) * | 2018-06-14 | 2024-02-20 | 腾讯科技（深圳）有限公司 | Method and device for processing operation data, computer readable medium and electronic equipment |
| CN109636577A (en) * | 2018-10-25 | 2019-04-16 | 深圳壹账通智能科技有限公司 | IP address analysis method, device, equipment and computer readable storage medium |
| CN109976934A (en) * | 2019-02-26 | 2019-07-05 | 上海拍拍贷金融信息服务有限公司 | Equipment uses abnormality determination method, device and computer storage medium |
| CN109976934B (en) * | 2019-02-26 | 2024-03-08 | 上海拍分乐网络科技有限公司 | Device use abnormality determination method, device and computer storage medium |
| CN110414914A (en) * | 2019-06-17 | 2019-11-05 | 深圳壹账通智能科技有限公司 | Business datum monitoring method and device |
| WO2020253395A1 (en) * | 2019-06-17 | 2020-12-24 | 深圳壹账通智能科技有限公司 | Service data monitoring method and apparatus |
| CN113313572A (en) * | 2021-05-28 | 2021-08-27 | 上海浦东发展银行股份有限公司 | Model identification method based on accumulation fund point-credit customer |
| CN113313572B (en) * | 2021-05-28 | 2022-12-20 | 上海浦东发展银行股份有限公司 | Model identification method based on accumulation fund point-credit customer |
| CN114693188A (en) * | 2022-05-31 | 2022-07-01 | 四川骏逸富顿科技有限公司 | Risk supervision method, system and equipment for drug retail industry |
| CN117708183A (en) * | 2023-11-08 | 2024-03-15 | 广州西米科技有限公司 | Potential user mining method and system based on user consumption habit |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107093101A (en) | Potential loan usage mining and risk score method based on POS pipelined datas | |
| Sitompul et al. | The Influence of Exchange Rate, Inflation, For the Results of the Development Assets of Islamic Banks | |
| US20210166167A1 (en) | Artificial intelligence and blockchain-based inter-enterprise credit rating and risk assessment method and system | |
| CN111476660B (en) | Intelligent wind control system and method based on data analysis | |
| EP1359525A1 (en) | Device for integrating transaction information on finantial transaction | |
| Ulansari et al. | A comparative study of the efficiency of conventional and Sharia insurance in Indonesia | |
| KR101686305B1 (en) | Apparatus for evaluating mortage loan through local risk measurement | |
| Wang et al. | Using small business banking data for explainable credit risk scoring | |
| Orok et al. | Inflation and deposit mobilization in deposit money banks: the Nigerian perspective | |
| CN105427171A (en) | Data processing method of Internet lending platform rating | |
| Cici et al. | Explaining and benchmarking corporate bond returns | |
| TWI770426B (en) | Cross-enterprise credit rating and risk assessment system | |
| Tran | Uncertainty and Investment: Evidence from A ustralian Firm Panel Data | |
| US8626633B1 (en) | Website and computer program for transfer of return on investment on a real time basis for OiBiTDA | |
| CN115345745A (en) | Financial wind control system for gold and jewelry industry and operation method | |
| US20030018548A1 (en) | System for economic-financial planning | |
| CN112884259A (en) | Cross-enterprise credit rating and risk assessment method and system | |
| JP2012238073A (en) | Credit purchase assessment support system and credit purchase assessment support method | |
| US20140278751A1 (en) | System and method for identifying rapidly-growing business customers | |
| Muflikhah et al. | The Effect Of Fee-Based Income, Third Party Funds, And Profit Sharing Rates On The Net Profit Of Islamic Banks For The 2016-2020 Period | |
| Machmud | Analysis of structure, conduct and performance of sharia insurance industry in Indonesia | |
| Rizkyanto et al. | Customer Segmentation of Personal Credit using Recency, Frequency, Monetary (RFM) and K-means on Financial Industry | |
| Péter Kolozsi et al. | Interbank liquidity and short-term yields in an emerging market economy–the experience of Hungary in 2016–2020 | |
| Niu et al. | Research on the Debt Default Risk of Guangyi Technology | |
| CN115423631A (en) | Trading member scoring method and system based on trading data of industrial internet platform |
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: 20170825 |