CN108460674A - Information processing method, device, computer equipment and storage medium - Google Patents
Information processing method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN108460674A CN108460674A CN201810102329.7A CN201810102329A CN108460674A CN 108460674 A CN108460674 A CN 108460674A CN 201810102329 A CN201810102329 A CN 201810102329A CN 108460674 A CN108460674 A CN 108460674A
- Authority
- CN
- China
- Prior art keywords
- credit rating
- model
- active user
- consumption
- obtains
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
Abstract
This application involves a kind of information processing method, device, computer equipment and storage mediums.The method includes:Obtain active user first time period multiple first position data and the active user second time period multiple second position data;The corresponding consumption data of each first position data is obtained, corresponding consumption feature is obtained according to each consumption data;The consumption feature is inputted in the credit rating model that training obtains in advance, obtains the corresponding target credit rating of the active user;The second position data are compared with predeterminated position data, obtain abnormal position comparison result;The corresponding resource transfers strategy of the active user is obtained according to the abnormal position comparison result and the target credit rating.The accuracy of resource numerical value transition strategy can be improved using this method.
Description
Technical field
This application involves technical field of information processing, more particularly to a kind of information processing method, device, computer equipment
And storage medium.
Background technology
With the development of Internet technology, people are more and more frequent to the use of internet, and more and more users pass through
Network carries out resource numerical value transfer.Therefore all there is the demand for determining resource transfers strategy under many scenes, such as in credit
In field, generally require according to user information such as age, income the really amount offered loans of directional user, at present mainly according to
The information filled in by user oneself carries out resource transfers Policy evaluation, however the information verification difficulty that user submits is larger, is not allowed
Really, the accuracy for leading to obtain resource numerical value transition strategy is low.
Invention content
Based on this, it is necessary to which in view of the above technical problems, providing a kind of can improve to obtain resource numerical value transition strategy
Information processing method, device, computer equipment and the storage medium of accuracy.
A kind of information processing method, the method includes:Active user is obtained in multiple first positions of first time period
The multiple second position data of data and the active user in second time period;Obtain each first position data pair
The consumption data answered obtains corresponding consumption feature according to each consumption data;The consumption feature is inputted into instruction in advance
In the credit rating model got, the corresponding target credit rating of the active user is obtained;By the second position data and in advance
If position data is compared, abnormal position comparison result is obtained;According to the abnormal position comparison result and the target
Credit rating obtains the corresponding resource transfers strategy of the active user.
In one embodiment, described to input the consumption feature in the credit rating model that training obtains in advance, it obtains
The step of target credit rating of the active user includes:The consumption feature is separately input to what multiple advance training obtained
In the credit rating model, the initial user's credit of the active user of each credit rating model output is obtained;According to institute
The weight calculation for stating each initial user's credit and corresponding credit rating model obtains the corresponding target credit of the active user
Degree.
In one embodiment, the method further includes:The sample set for carrying out model training is obtained, the sample set includes
Multiple samples, the sample include multiple trained consumption features and corresponding sample credit rating;According to the sample set and
A variety of different model training methods carry out model trainings, obtain each different model training method train it is more
A credit rating model;The corresponding consumption feature of the sample is input in the credit rating model, the sample is obtained and corresponds to
Model credit rating;According to the difference of sample corresponding model credit rating and the sample credit rating in each credit rating model
Away from obtaining the corresponding weight of each credit rating model.
In one embodiment, described to be believed according to the corresponding model credit rating of sample in each model and the sample
The gap of expenditure obtains the step of each model corresponding weight and includes:Calculate sample pair in each credit rating model
The deviation of the model credit rating and the sample credit rating answered;Summation meter is carried out to the corresponding deviation of each credit rating model
It calculates, obtains the corresponding total deviation of each credit rating model;According to the corresponding total deviation of each credit rating model and
Preset Weight algorithm obtains the corresponding weight of each credit rating model, wherein total deviation and power in the Weight algorithm
Weight is negative correlativing relation.
In one embodiment, described to obtain the corresponding consumption data of each first position data, according to described each
The step of a consumption data obtains corresponding consumption feature include:Obtain the corresponding trade company's letter of each first position data
Breath, the merchant information includes the consumption data of the trade company and the merchant type of the trade company;According to each trade company
Consumption data and the consumption type of the trade company obtain corresponding consumption feature.
In one embodiment, the predeterminated position data include historical actions of the active user in the third period
Track, the second position data include the active user in the goal activity track of second time period, described by described
Two position datas are compared with predeterminated position data, and the step of obtaining abnormal position comparison result includes:It obtains described current
User is in the time where the historical actions track of the third period, the third period earlier than the second time period
The time at place;By the active user in the goal activity track of the second time period and the active user described the
The historical actions track of three periods is compared, and abnormal position comparison result is obtained.
In one embodiment, it is described obtained according to the abnormal position comparison result and the target credit rating it is described
The step of the corresponding resource transfers strategy of active user includes:It is corresponding that the active user is obtained according to the target credit rating
Initial resource shifts numerical value;The behavior pattern of the active user is determined according to the abnormal position comparison result;According to described
Behavior pattern and initial resource transfer numerical value obtain the corresponding target resource transfer numerical value of the active user.
A kind of information processing unit, described device include:Position data acquisition module, for obtaining active user first
The multiple second position data of multiple first position data of period and the active user in second time period;Feature obtains
To module, for obtaining the corresponding consumption data of each first position data, obtained pair according to each consumption data
The consumption feature answered;Credit rating obtains module, for the consumption feature to be inputted in the credit rating model that training obtains in advance,
Obtain the corresponding target credit rating of the active user;Contrast module is used for the second position data and predeterminated position number
According to being compared, abnormal position comparison result is obtained;Strategy obtains module, for according to the abnormal position comparison result and
The target credit rating obtains the corresponding resource transfers strategy of the active user.
In one embodiment, the credit rating obtains module and includes:Initial user's credit obtains unit, for disappearing described
Expense feature is separately input in the credit rating model that multiple advance training obtain, and obtains each credit rating model output
The active user initial user's credit;Target credit rating obtains unit, for according to each initial user's credit and
The weight calculation of corresponding credit rating model obtains the corresponding target credit rating of the active user.
In one embodiment, described device further includes:Sample set acquisition module, for obtaining the sample for carrying out model training
This collection, the sample set include multiple samples, and the sample includes multiple trained consumption features and corresponding sample credit rating;
Training module obtains described each for carrying out model training according to the sample set and a variety of different model training methods
Multiple credit rating models that a different model training method is trained;Model credit rating module is used for the sample pair
The consumption feature answered is input in the credit rating model, obtains the corresponding model credit rating of the sample;Weight obtains module,
For obtaining institute according to the gap of the corresponding model credit rating of sample in each credit rating model and the sample credit rating
State the corresponding weight of each credit rating model.
In one embodiment, the weight obtains module and is used for:Sample in each credit rating model is calculated to correspond to
Model credit rating and the sample credit rating deviation;Summation meter is carried out to the corresponding deviation of each credit rating model
It calculates, obtains the corresponding total deviation of each credit rating model;According to the corresponding total deviation of each credit rating model and
Preset Weight algorithm obtains the corresponding weight of each credit rating model, wherein total deviation and power in the Weight algorithm
Weight is negative correlativing relation.
In one embodiment, the feature obtains module and includes:Merchant information acquiring unit, it is described each for obtaining
The corresponding merchant information of first position data, the merchant information include the consumption data of the trade company and the quotient of the trade company
Family type;Feature obtains unit, for being obtained according to the consumption data of each trade company and the consumption type of the trade company
Corresponding consumption feature.
In one embodiment, the predeterminated position data include historical actions of the active user in the third period
Track, the contrast module include:Historical track acquiring unit, for obtaining the active user in the third period
Historical actions track, the time where the third period is earlier than the time where the second time period;Comparing unit is used
In by the active user in the goal activity track of the second time period and the active user in the third period
Historical actions track compared, obtain abnormal position comparison result.
In one embodiment, the strategy obtains module and includes:Initial value obtains unit, for according to the target
Credit rating obtains the corresponding initial resource transfer numerical value of the active user;Behavior pattern determination unit, for according to described different
Normal position comparison result determines the behavior pattern of the active user;Target value obtains unit, for according to the behavior mould
Formula and initial resource transfer numerical value obtain the corresponding target resource transfer numerical value of the active user.
A kind of computer equipment, including memory, processor and storage can be run on a memory and on a processor
Computer program, the processor realize following steps when executing the computer program:Active user is obtained at the first time
Section multiple first position data and the active user second time period multiple second position data;It obtains described each
The corresponding consumption data of a first position data, corresponding consumption feature is obtained according to each consumption data;Disappear described
In the expense feature input credit rating model that training obtains in advance, the corresponding target credit rating of the active user is obtained;It will be described
Second position data are compared with predeterminated position data, obtain abnormal position comparison result;It is compared according to the abnormal position
As a result and the target credit rating obtains the corresponding resource transfers strategy of the active user.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
Following steps are realized when row:Active user is obtained in multiple first position data of first time period and the active user to exist
Multiple second position data of second time period;The corresponding consumption data of each first position data is obtained, according to described
Each consumption data obtains corresponding consumption feature;The consumption feature is inputted in the credit rating model that training obtains in advance,
Obtain the corresponding target credit rating of the active user;The second position data are compared with predeterminated position data, are obtained
To abnormal position comparison result;The active user is obtained according to the abnormal position comparison result and the target credit rating
Corresponding resource transfers strategy.
Above- mentioned information processing method, device, computer equipment and storage medium, by obtaining active user at the first time
Multiple first position data of section and active user obtain each first in multiple second position data of second time period
The corresponding consumption data of data is set, corresponding consumption feature is obtained according to each consumption data, consumption feature is inputted into instruction in advance
In the model got, the target credit rating of active user is obtained, second position data and predeterminated position data are compared,
Abnormal position comparison result is obtained, the corresponding resource of active user is obtained according to abnormal position comparison result and target credit rating
Transition strategy.Since user location can be utilized to obtain the credit rating of user, and further combined with the exception that user location obtains
Position comparison result obtains resource transfers strategy, therefore obtained resource transfers strategy accuracy is high.
Description of the drawings
Fig. 1 is the application scenario diagram of information processing method in one embodiment;
Fig. 2 is the flow diagram of information processing method in one embodiment;
Fig. 3 is to input consumption feature in the credit rating model that training obtains in advance in one embodiment, is currently used
The flow diagram of the step of credit rating of the corresponding active user in family;
Fig. 4 is the flow diagram of information processing method in one embodiment;
Fig. 5 is to obtain the corresponding consumption data of each first position data in one embodiment, according to each consumption data
The flow diagram for the step of obtaining corresponding consumption feature;
Fig. 6 is corresponding to obtain active user according to abnormal position comparison result and target credit rating in one embodiment
The flow diagram of the step of resource transfers strategy;
Fig. 7 is to compare second position data and predeterminated position data in one embodiment, obtains abnormal position ratio
The flow diagram of the step of to result;
Fig. 8 is the structure diagram of information processing unit in one embodiment;
Fig. 9 is that credit rating obtains the structure diagram of module in one embodiment;
Figure 10 is the structure diagram of information processing unit in one embodiment;
Figure 11 is the internal structure chart of one embodiment Computer equipment.
Specific implementation mode
It is with reference to the accompanying drawings and embodiments, right in order to make the object, technical solution and advantage of the application be more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
Information processing method provided by the present application can be applied in application environment as shown in Figure 1.Wherein, terminal 102
It is communicated by network with server 104 by network.Terminal 102 is the terminal that active user logs in, for example, pacifying in terminal
Equipped with the application for carrying out resource transfers, active user can log in the application, and application can obtain the location information of present terminal
Position data as active user is sent in server 104, terminal 102 can in real time, it is random or the set time to
Server 104 sends position data and is protected to position data after server 104 receives the position data of the transmission of terminal 102
It deposits, when needing to obtain the corresponding resource transfers strategy of active user for example when the loan requests for receiving active user's transmission
Afterwards, server 104 can obtain terminal 102 when the multiple first position data and terminal 102 of first time period are second
Between section multiple second position data, to obtain the corresponding resource transfers strategy of active user.Wherein, terminal 102 can with but not
It is limited to various personal computers, laptop, smart mobile phone, tablet computer and portable wearable device, server 104
It can be realized with the server cluster of the either multiple server compositions of independent server.
In one embodiment, as shown in Fig. 2, providing a kind of information processing method, it is applied in Fig. 1 in this way
It illustrates, includes the following steps for server:
Step S202 obtains active user in multiple first position data of first time period and active user second
Multiple second position data of period.
Specifically, the number of first position data and second position data can be configured as needed, first
The number for setting data and second position data can be the same or different, and first time period and second time period can be identical
It can also be different, position data can be that the corresponding present terminal of active user is sent in server, can also be server
It is obtained from the equipment of the position data of other storages active user, the embodiment of the present invention obtains active user couple to server
The source for the position data answered is not limited.For example, position data conduct of the active user within past one month can be obtained
First position data obtain position data of the active user within past two weeks as second position data.Or it can obtain
Active user, as first position data, will use in the position data for consuming place place such as market, entertainment venues, food and drink
Family house, office position data as second position data.
In one embodiment, present terminal can to server send GPS (Global Positioning System,
Global positioning system) coordinate information, active user user identifier, therefore, after server receives GPS information, by GPS information
As the corresponding position data of active user.In some embodiments, present terminal can send active user's to server
Location dependent information, such as can be the mark of signal strength information and signal transmitting equipment that signal transmitting equipment is sent,
Therefore server can calculate to obtain according to signal strength present terminal at a distance from signal transmitting equipment, then further according to signal
The position and present terminal for emitting equipment obtain the position of active user at a distance from signal transmitting equipment.Signal emits equipment
Such as can be WiFi (Wireless Fidelity, Wireless Fidelity) equipment, bluetooth equipment, Zigbee equipment etc..
In one embodiment, if obtaining less than sufficiently accurate position, it can only obtain the position area where active user
When domain, then multiple position area informations of active user can be obtained, are then used according to the position relationship between the band of position
The exact position at family.In one embodiment, three bands of position can be obtained, the coincidence area of three bands of position is then calculated
Domain obtains corresponding location information according to overlapping region.For example, when there are overlapping region in 3 bands of position, then weight can be obtained
The region crosspoint between the boundary of three bands of position respectively is closed, with the triangle of the mutual line formation of 3 crosspoints
Position of the heart coordinate as active user.When 3 bands of position are intersected two-by-two, then intersection two-by-two can be obtained respectively and is obtained
Overlapping region and the band of position boundary intersection point, using the midpoint of the line of intersection point as vertex of a triangle, then by three
Position of the angular heart as active user.It is appreciated that can then obtain other 3 when three regions are not intersected
The band of position.If continuous three bands of position obtained three times are not intersected, last trizonal center can be utilized
It is formed by position of the heart as active user of triangle.
In one embodiment, the algorithm for the position data for obtaining active user using three bands of position is specific as follows:
Obtain the central point of these three bands of position, it is assumed that be (X1, Y1) that the radius of (X2, Y2), (X3, Y3), each band of position are
Then d1, d2 and d3 construct this corresponding equation in 3 bands of position:(x-Xi) 2+ (y-Yi) 2=di, wherein Xi are referred to
The abscissa of i-th of band of position, Yi refer to that the ordinate of i-th of band of position, di refer to the radius of i-th of band of position, i
1,2,3 can be taken.This 3 equations are subjected to combination of two, form three equation groups, three equation groups is solved and respectively obtains target
Coordinate points.If solve the obtained coordinates of targets point of equation group be less than or equal to 2, can continue to look for the next group of band of position or
Using the heart for the triangle of these three bands of position being centrally formed as the position of active user.If coordinates of targets point be more than etc.
In 3, then the corresponding region of triangle that every 3 coordinates of targets point is formed can be compared with 3 bands of position, if depositing
In the triangle for being respectively positioned on these three bands of position, then the heart of the triangle is the position of active user.If equal position is not present
Triangle in these three bands of position then obtains the midpoint of the line of two coordinates of targets points of each equation group, then with
The heart for the triangle that this 3 midpoints are formed is as the corresponding position of active user.
In one embodiment, three bands of position can be obtained by GPS information, can also obtain the IP of terminal
Address obtains corresponding base station location, then using the band of position where base station as the band of position of terminal according to IP address.
It is in one embodiment, for example indoor for that can not be accurately positioned to obtain the place of exact position residing for terminal by GPS,
It can be used by the position and present terminal received signal Strength co-mputation of the Zigbee devices on indoor such as market
Family exact position.
In one embodiment, due to needing lasting acquisition user location, in order to protect the privacy of user, present terminal
Obtained location information can be obscured by position Obfuscating Algorithms, therefore according to the position data that receives of server
Physical location obscured after position data, therefore server can carry out it is antialiasing processing obtain actual position data.Instead
Obscuring the method for processing can specifically be configured according to position Obfuscating Algorithms.Such as when Obfuscating Algorithms are practical longitude and latitude
Respectively plus at 2 degree, then antialiasing algorithm is respectively to subtract 2 by the longitude and latitude received degree obtains physical location.
Step S204 obtains the corresponding consumption data of each first position data, is corresponded to according to each consumption data
Consumption feature.
Specifically, after obtaining first position data, the consumption data in the place where each first position is obtained.Place
Place can be the places such as market, tourism, lodging, amusement, and the consumption data in each place can be the pre-capita consumption in the place
For example, 800 yuan, can also be indicated using consumption grade, such as consumption data can be the place consumption levels be it is high,
In, it is low.It can be configured according to actual needs using the method that consumption data obtains corresponding consumption feature, in a reality
It applies in example, a consumption feature can be obtained according to each corresponding consumption data of first position data, in one embodiment
In a consumption feature can also be obtained according to multiple consumption datas for example by corresponding consumption data at various locations daily
With obtain a consumption feature.The trade company where each first position can also be classified in one embodiment, such as
It is divided into existence class, development class and enjoys class.Then the level of consumption of each merchant type is counted as generated class trade company per capita
Consumption is 60 yuan, obtains the corresponding consumption feature of user.Active user can also be counted into each merchant type and corresponding
The number of the trade company of the level of consumption, it is such as average monthly to enter life kind trade company and time of trade company of the level of consumption for 200~500 yuan
Number is 3 times, obtains corresponding one consumption feature, the average life kind trade company and the level of consumption of monthly entering is 1000~1500 yuan
Trade company number be 1 time, obtain another consumption feature.The rule that corresponding consumption feature is obtained according to consumption data is specific
It can be configured according to actual needs.For example, characteristic vector space can be mapped to obtained consumption data, such as to life
The number for the trade company that class trade company and the level of consumption are 200~500 yuan is 3 corresponding consumption features
[10000000000], the behavior that the number for the trade company for being 1000~1500 yuan into life kind trade company and the level of consumption is 1 time is special
Sign can be expressed as [010000000000], and the dimension of consumption feature can be specifically configured according to actual needs.Such as 50
Dimension etc..
Consumption feature is inputted in the credit rating model that training obtains in advance, it is corresponding to obtain active user by step S206
The target credit rating of active user.
Specifically, credit rating is used to weigh the degree of user credit height, and credit rating can use specific numerical tabular example
It if credit score is indicated, can also be indicated by rank, such as high, medium and low etc..Consumption feature is input to advance instruction
In experienced credit rating model, corresponding target credit rating can be obtained.Credit rating model can be one or more.It is multiple when having
When, target credit rating can be obtained according to the credit rating that multiple credit rating models obtain, such as can be by multiple credit rating models
Average value, median, highest credit rating or the minimum credit rating of obtained credit rating can also be obtained as target credit rating
The corresponding weight of each credit rating model, the credit rating obtained according to the weight of credit rating model and each credit rating model obtain
To target credit rating.The weight of credit rating model can rule of thumb, need be arranged, the effect of model training can also be passed through
Fruit obtains.For example, when there are three credit rating model, the corresponding weight of each credit rating model can be respectively set to 0.5,
0.3,0.2, if the corresponding credit score of each credit rating model is respectively 600,500,800, target credit rating=0.5*600
+ 0.3*500+0.2*800=610 points.
Credit rating model carries out model training previously according to training data and obtains.Model instruction is carried out by training data
Practice, can determine the corresponding model parameter of each consumption feature, to which the model parameter obtained according to training obtains credit rating mould
Type.When carrying out model training, may be used the model training mode of supervision, for example, Logic Regression Models, Bayesian model,
Adaptive algorithm, SVM (Support Vector Machine, support vector machines) etc..For example, known credit rating can be obtained
And corresponding consumption feature, consumption feature and known credit rating are then subjected to model training as training data.With
For SVM, stochastic gradient descent algorithm may be used in the training process and carry out model training, needed in gradient descent procedures
So that the minimum corresponding model parameters of cost function J (θ), to obtain credit rating model.
Step S208 compares second position data and predeterminated position data, obtains abnormal position comparison result.
Specifically, abnormal comparison result includes existing abnormal or being not present abnormal.Predeterminated position data can be current
Home address, office location of user etc., predeterminated position data can be that active user fills in, and can also be to be used according to current
The daily location information in family can for example obtain the historical track of user, the family of user is obtained according to the rule of historical track
Location, business address etc..Second position data and predeterminated position data are compared, are obtained with the presence or absence of abnormal criterion
It can be configured as needed, such as the movement track of active user is obtained for example in the second time according to second position data
The movement track of every day in section, home address, business address or the past day for then filling in movement track and user
Normal movement track is compared, and judges to be whether the different movement tracks that whether there is exception.
Step S210 obtains the corresponding resource transfers of active user according to abnormal position comparison result and target credit rating
Strategy.
Specifically, it can be loan that resource can go to another account for example from an account.Resource transfers strategy packet
Resource transfers numerical value is included, can also include resource transfers condition in one embodiment, resource transfers condition for example may include
Guarantee condition, interest provided a loan etc..It needs to consider abnormal position comparison result and the target letter of active user
Expenditure obtains resource transfers strategy.In one embodiment, the resource transfers of active user can be obtained according to target credit rating
Numerical value obtains the interest or needs that active user carries out the condition such as loan of resource transfers according to abnormal position comparison result
Guarantee condition of offer etc., if the quantity of abnormal position is more or deposits when abnormal, corresponding condition requires high.In a reality
It applies in example, the resource transfers numerical value of active user can be obtained according to target credit rating, determined according to abnormal position comparison result
Whether the interim request that improves resource transfers numerical value of active user proposition is met.
In one embodiment, on a timeline, the corresponding initial time of first time period is corresponded to earlier than second time period
Initial time, in one embodiment, the time span of first time period can also be more than the time span of second time period,
Such as second time period can be that recent data for example be pass by one month or bimestrial data, first time period can be with
For 1 year data of past.Therefore, determine that the credit rating of user obtains the resource transfers number of user using long-term consumption feature
The accuracy of value is high, and then can be used for illustrating currently using with the presence or absence of abnormal position comparison result in short-term second time period
Family whether there is exception in the recent period, therefore can be adjusted to resource transfers strategy using abnormal position comparison result, to improve money
The accuracy of source transition strategy.If for example, determining that the continuous surrounding of user does not reach Office Area by abnormal position comparison result
Domain then can tentatively judge that active user is unemployed, and when resource is loan, can reduce the loan obtained according to target credit rating
Amount to reduce risk, or improves guarantee condition.
In one embodiment, obtained resource transfers strategy can also be pushed on the present terminal of active user.
Can be sent when the resource transfers for receiving present terminal transmission ask such as loan requests, can also be server master
Dynamic push, it is not limited herein.
In above- mentioned information processing method, by obtain active user first time period multiple first position data and
Active user obtains the corresponding consumption data of each first position data, root in multiple second position data of second time period
Corresponding consumption feature is obtained according to each consumption data, consumption feature is inputted in the model that training obtains in advance, is obtained current
The target credit rating of user, second position data and predeterminated position data are compared, and obtain abnormal position comparison result, root
The corresponding resource transfers strategy of active user is obtained according to abnormal position comparison result and target credit rating.Since use can be utilized
Family position obtains the credit rating of user, and the resource that the abnormal position comparison result obtained further combined with user location obtains turns
Strategy is moved, therefore the accuracy for obtaining resource transfers strategy is high, can also reduce the push of invalid resource transfers strategy, improves
The utilization rate of computer network resources.
In one embodiment, as shown in figure 3, step S206 is that consumption feature is inputted the credit rating that training obtains in advance
In model, the step of obtaining the credit rating of the corresponding active user of active user, includes:
Consumption feature is separately input in multiple credit rating models trained in advance, obtains each credit by step S302
Spend the initial user's credit of the active user of model output.
Specifically, the quantity of credit rating model can be configured according to actual needs, such as can be 3.Disappeared
After taking feature, consumption feature is separately input in the credit rating model that training obtains in advance, it is defeated to obtain each credit rating model
The initial user's credit of the active user gone out.
Step S304 obtains active user according to the weight calculation of initial user's credit and corresponding credit rating model and corresponds to
Target credit rating.
Specifically, the corresponding weight of each credit rating model can be self-defined setting as needed, can also basis
The accuracy rate of model determines the corresponding weight of each credit rating model when carrying out model training.After obtaining initial user's credit, according to
Each initial user's credit and the corresponding weight of corresponding credit rating model obtain the corresponding target credit rating of active user.Example
Such as, when there are four credit rating model, the corresponding weight of each credit rating model is respectively 0.4,0.3,0.2,0.1, Ge Gexin
Expenditure model output initial credit grade be respectively it is high, high and low, in, since the weight that credit grade is high model is
0.7, it is more than the weight of other grades, then target credit rating is height.
As shown in figure 4, in one embodiment, information processing method further includes:
Step S402 obtains the sample set for carrying out model training, and sample set includes multiple samples, and sample includes multiple training
Consumption feature and corresponding sample credit rating.
Specifically, the number of samples in sample set can be arranged as required to or randomly select, such as can be 100,000
It is a, the training consumption feature in each sample can be obtained according to the consumption data of the position of multiple training users, such as
The consumption data of the trade company of position where multiple training users obtains the consumption feature of training user.Sample credit rating can
To be manually to mark, can also be recorded according to the reference of user by other channels such as bank.Sample is used for mould
Type is trained, and credit rating model is obtained with training.
Step S404 carries out model training according to sample set and a variety of different model training methods, obtain it is each not
Multiple credit rating models that same model training method is trained.
Specifically, different model training methods can refer to the process of the model difference used or training difference etc..
For example, SVM is respectively adopted, neural network is different model training method.When using SVM, if the kernel function used is different,
Also it is different model training method.After obtaining sample set, carried out using sample set and a variety of different model training methods
Model training obtains a variety of models.During carrying out model training, since sample credit rating is known and to there is supervision
Model, therefore can be obtained according to the consumption feature and model parameter of the sample of input by constantly adjusting model parameter
To credit rating meet practical or close to known sample credit rating, the model parameter obtained so as to basis obtains
Credit rating model.The model of model training can be SVM (Support Vector Machine, support vector machines) grader mould
Type, neural network (Artificial Neural Network, ANN) sorter model, logistic regression algorithm (logistic
Regression, LR) sorter model and hidden Markov model (Hidden Markov Model, HMM) etc. are various can be into
The model of row machine learning.
The corresponding consumption feature of sample is input in credit rating model by step S406, obtains the corresponding model letter of sample
Expenditure.
Specifically, when training obtains each credit rating model, the consumption feature of sample in sample set is input to respectively
In trained credit rating model, obtains each sample and input the model credit rating exported after trained credit rating model.
Step S408 is obtained each according to the gap of the corresponding model credit rating of sample in each model and sample credit rating
The corresponding weight of credit rating model.
Specifically, after the model credit rating for obtaining sample in each model, computation model credit rating and sample credit rating
Gap, to obtain the corresponding weight of each credit rating model according to gap.For example, it is assumed that the model credit rating of a samples is 80 points,
Sample credit rating when a samples carry out model training is 90 points, then the gap of model credit rating and sample credit rating is 10 points.Again
For example, it is 0.8 that the credit grade that a samples are exported in B credit rating models, which is high probability, and the sample credit rating in sample is
Height, i.e. probability are 1.Then a samples are 0.2 in the deviation of b models.According to the corresponding model credit rating of sample in model and sample
The gap of credit rating obtains the corresponding weight of each credit rating model can specifically be configured according to actual needs.In a reality
It applies in example, the gap of the model credit rating and sample credit rating of each sample in each credit rating model can be calculated
With the then weight corresponding with credit rating model is obtained according to gap.Gap and with the negatively correlated relationship of weight.For example,
The weight of credit rating model with for the corresponding gap of the credit rating model and inverse.In some embodiments, it obtains each
The corresponding gap of credit rating model and after, can to the corresponding gap of each model and be normalized, after normalization
Value as corresponding weight.
In one embodiment, it is obtained according to the corresponding model credit rating of sample in each model and the gap of sample credit rating
The step of weight corresponding to each model may include:Calculate in each credit rating model the corresponding model credit rating of sample with
The deviation of sample credit rating carries out read group total to the corresponding deviation of each credit rating model, obtains each credit rating model pair
The total deviation answered obtains each credit rating model according to the corresponding total deviation of each credit rating model and preset Weight algorithm
Corresponding weight, wherein total deviation and weight are negative correlativing relation in Weight algorithm.
Specifically, if to refer to total deviation big for negative correlativing relation, weight is small, if total deviation is small, weight is big.If for example,
The total deviation of one model is 90, and the total deviation of the second model is 100.The weight ratio of the first model then obtained according to Weight algorithm
The weight of second model is big.Weight algorithm can be configured according to actual needs.Such as can be linear function, or
Exponential function.
For example, it is assumed that there are three samples for sample set:A samples, B samples, C samples.And obtain two according to three sample trainings
A credit rating model:A samples, B samples, C samples are separately input to train in advance first by the first model and the second model
In model and the second model, obtain model credit rating a1, b1 that A samples, B samples, C samples are exported in the first model and
The model credit rating that c1, A sample, B samples, C samples are exported in the second model is respectively a2, b2 and c2.Obtain model output
Model credit rating after, calculate A samples credit rating and a1, A sample credit rating and a2, B sample credit rating and b1, B sample credit
Degree and the deviation of b2, C sample credit rating and c1, C sample credit rating and c2, it is assumed that be a11, a21, b11, b12, c11 and
c12.Then the corresponding model bias value of the first model carries out read group total and obtains the total deviation of the first model to be a11+b11+c11,
The corresponding total deviation value a21+b21+c21 of second model.Then model pair is used as after the inverse of total deviation value being normalized
The weight answered.
As shown in figure 5, in one embodiment, the corresponding consumption data of each first position data is obtained, according to each
Consumption data obtains the step of corresponding consumption feature and includes:
Step S502, obtains the corresponding merchant information of each first position data, and merchant information includes the consumption number of trade company
According to this and the merchant type of trade company.
Specifically, after obtaining first position data, the trade company obtained on first position is corresponding as first position data
Merchant information, trade company can be food and drink trade company, amusement trade company or trade company of hotel etc..Consumption data can be disappearing per capita for trade company
It is, for example, 800 yuan to take, and can also be to be indicated using the consumption grade of the trade company, such as consumption data can be the consumption of the trade company
Rank is high, medium and low.Merchant type can be classified to obtain as needed.Such as it can be life kind, develop class and enjoy
By class etc..Life kind can be supermarket, development class for example can be all kinds of Training and Learning schools, enjoy class can be tourism scape
Point, Entertainment Scene are such as KTV.
Step S504 obtains corresponding consumption feature according to the consumption type of the consumption data of each trade company and trade company.
Specifically, the number of consumption feature can be arranged as required to, such as can be 50.Obtain consumption data and
After the consumption type of trade company, the corresponding consumption data of trade company of each consumption type can be counted, is corresponded to according to consumption type
Consumption data obtain consumption feature, such as count the average per capita consumption of the trade company of each consumption type or in each consumption
The ratio of horizontal trade company, obtains corresponding consumption feature, for example, it is assumed that there are three the trade companies of life kind consumption type, consumption
Level is respectively 800,900 and 1000 yuan, then the average per capita consumption of the trade company of life kind consumption type can be calculated
For (800+900+1000)/3=800 members, it is assumed that the trade company of development class consumption type has 5, the level of consumption is respectively 500,
900,1500,5500 and 8000 yuan, then it is 0~1000 yuan that can calculate the trade company for developing class consumption type in the level of consumption
Ratio be 2/5*100%=40%, the level of consumption is that 1001 yuan~3000 yuan of ratio is 1/5*100%=20%, consumption
Level is 2/5*100%=40% for 3001 yuan or more of ratio.In one embodiment, active user's entrance can also be counted
The number of the trade company of each merchant type and the corresponding level of consumption such as averagely monthly enters life kind trade company and the level of consumption is
The number of 200~500 yuan of trade company is 3 times, obtains corresponding consumption feature.
In one embodiment, as shown in fig. 6, step S210 is i.e. according to abnormal position comparison result and target credit rating
The step for obtaining the corresponding resource transfers strategy of active user includes:
Step S602 obtains the corresponding initial resource of active user according to target credit rating and shifts numerical value.
Specifically, it is provided with the correspondence of credit rating and resource transfers numerical value, if credit rating is 90 points of corresponding resources
It is 9000 yuan to shift numerical value.Can also in conjunction with other factors such as the taking in of user, Zi An institutes are determined current debt situation
Shift numerical value, wherein income can be positive correlation with resource transfers numerical value, and debt number can be with resource transfers numerical value
Negative correlativing relation.
Step S604 determines the behavior pattern of active user according to abnormal position comparison result.
Specifically, behavior pattern may include one or more etc. in normal, unemployment, separation, divorce, excess consumption.
The method that the behavior pattern of active user is determined according to abnormal position comparison result can be arranged as required to, for example, when current
The abnormal position comparison result of user is that daily starting point is identical, when terminal difference, it is determined that user is unemployed.Or work as abnormal position
The level of consumption of trade company of the comparison result where position is excess consumption higher than the ability to bear of user.Or when user is every
The start position that its morning sets out is different, but final position is identical, then behavior pattern can be separation.
Step S606 obtains the corresponding target resource of active user according to behavior pattern and initial resource transfer numerical value and turns
Move numerical value.
Specifically, the corresponding influence numerical value of each behavior pattern can be set, initial resource is then shifted into numerical value and shadow
It rings numerical value to be added, obtains target resource transfer numerical value, for example, the corresponding influence numerical value of unemployment is -9000, separation is corresponding
It is -6000 to influence numerical value, and corresponding influence numerical value of divorcing is -10000.The corresponding influence numerical value of normal behaviour pattern is 0.Also may be used
The corresponding abnormality score of each behavior pattern is arranged, such as 80 points of unemployment is lived apart 50 points, and excess consumes 40 points, obtains behavior mould
After formula, abnormality score is counted, obtains abnormal total score, corresponding total influence numerical value is then obtained according to abnormal total score.Root
The corresponding target resource of active user, which is obtained, according to total influence numerical value and initial resource transfer numerical value shifts numerical value.
In one embodiment, predeterminated position data include active user in the historical actions track of third period, and
Two position datas include active user in the goal activity track of second time period.As shown in fig. 7, step S208 is i.e. by second
Setting the step of data are compared with predeterminated position data, obtain abnormal position comparison result includes:
Step S702 obtains active user in the historical actions track of third period, the time where the third period
Earlier than the time where second time period.
Specifically, the third period refers to existing time included by the third period earlier than the time where second time period
It is generally more early than second time period.The third period may include all or part of second time period in one embodiment,
The third period can also be the period to connect with second time period, for example, the third period is 2 months to August 2017,
Second time period be in September, 2017 so far.For example, the third period can be the data in past 1 year, second time period can
To be the data in past one month.Movement track refer to by route, the historical actions track of third period can have more
It plants, such as can be movement track, movement track weekly, workaday movement track or section daily in the third period
One or more among the periodics such as the movement track of holiday.
Step S704, by active user in the goal activity track of second time period and active user in the third period
Historical actions track is compared, and abnormal position comparison result is obtained.
Specifically, after obtaining historical actions track, goal activity track and historical actions track are compared, to confirm
Whether track is abnormal.When carrying out track comparison, if there are many types of the historical actions track of third period, it can be right
Each goal activity track and historical actions track are compared, and the comparison result of the track of each type is obtained, and are confirmed
The track of active user is with the presence or absence of abnormal.For example, by the festivals or holidays movement track and third period in second time period
Festivals or holidays movement track compared.By the working day movement track in second time period and the working day in the third period
Movement track is compared.Judge whether abnormal standard can be determined according to actual needs for track.In one embodiment
In, if the consumption data that the starting point of track is not all the trade company where exception or track is higher or lower than active user
Consuming capacity is abnormal.In one embodiment, can determine whether in conjunction with behavior pattern to be determined for abnormal position ratio
To result.When the behavior pattern to be determined be unemployment when, can be in second time period continuous 3 weeks working day movement tracks and
Working day movement track difference in the third period is abnormal, can also be starting point difference, but if terminal is same
Attribute is for example all that office building is then normal.It can be the working day in second time period for behavior pattern of living apart or divorce
Movement track is abnormal if starting point difference, can also further judge rising for the workaday track of the spouse of active user
Whether point is identical as the starting point of active user, if it is different, then position comparison result is abnormal.
It should be understood that although each step in above-mentioned flow chart is shown successively according to the instruction of arrow, this
A little steps are not that the inevitable sequence indicated according to arrow executes successively.Unless expressly state otherwise herein, these steps
It executes there is no the limitation of stringent sequence, these steps can execute in other order.Moreover, at least part among the above
Step may include that either these sub-steps of multiple stages or stage are executed in synchronization to multiple sub-steps
It completes, but can execute at different times, the execution sequence in these sub-steps or stage is also not necessarily to be carried out successively,
But it can either the sub-step of other steps or at least part in stage execute in turn or alternately with other steps.
In one embodiment, as shown in figure 8, providing a kind of information processing unit, including:Position data acquisition module
802, feature obtains module 804, credit rating obtains module 806, contrast module 808 and strategy obtain module 810, wherein:
Position data acquisition module 802, for obtain active user first time period multiple first position data with
And active user is in multiple second position data of second time period.
Feature obtains module 804, for obtaining the corresponding consumption data of each first position data, according to each consumption number
According to obtaining corresponding consumption feature.
Credit rating obtains module 806, for inputting consumption feature in the credit rating model that training obtains in advance, is worked as
The corresponding target credit rating of preceding user.
Contrast module 808 obtains abnormal position comparison for comparing second position data and predeterminated position data
As a result.
Strategy obtains module 810, for obtaining active user couple according to abnormal position comparison result and target credit rating
The resource transfers strategy answered.
In one embodiment, as shown in figure 9, credit rating obtains module 806 includes:
Initial user's credit obtains unit 806A, for consumption feature to be separately input to the credit that multiple advance training obtain
It spends in model, obtains the initial user's credit of the active user of each credit rating model output.
Target credit rating obtains unit 806B, for the power according to each initial user's credit and corresponding credit rating model
Re-computation obtains the corresponding target credit rating of active user.
In one embodiment, as shown in Figure 10, information processing unit further includes:
Sample set acquisition module 1002, for obtaining the sample set for carrying out model training, sample set includes multiple samples, sample
This includes multiple trained consumption features and corresponding sample credit rating.
Training module 1004 is obtained for carrying out model training according to sample set and a variety of different model training methods
The multiple credit rating models trained to each different model training method.
Model credit rating module 1006 obtains sample for the corresponding consumption feature of sample to be input in credit rating model
This corresponding model credit rating.
Weight obtains module 1008, for being believed according to the corresponding model credit rating of sample in each credit rating model and sample
The gap of expenditure obtains the corresponding weight of each credit rating model.
In one embodiment, weight obtains module 1008 and is used for:Calculate the corresponding mould of sample in each credit rating model
The deviation of type credit rating and sample credit rating.Read group total is carried out to the corresponding deviation of each credit rating model, obtains each letter
The corresponding total deviation of expenditure model.It is obtained according to the corresponding total deviation of each credit rating model and preset Weight algorithm each
The corresponding weight of credit rating model, wherein total deviation and weight are negative correlativing relation in Weight algorithm.
In one embodiment, feature obtains module 804 and includes:Merchant information acquiring unit, for obtaining each first
The corresponding merchant information of position data, merchant information include the consumption data of trade company and the merchant type of trade company.Feature obtains
Unit, for obtaining corresponding consumption feature according to the consumption data of each trade company and the consumption type of trade company.
In one embodiment, predeterminated position data include active user in the historical actions track of third period, and
Two position datas include active user in the goal activity track of second time period, and contrast module 808 includes:Historical track obtains
Unit, for obtaining active user in the historical actions track of third period, the time where the third period is earlier than second
Time where period.Comparing unit is used for active user in the goal activity track of second time period and active user
It is compared in the historical actions track of third period, obtains abnormal position comparison result.
In one embodiment, strategy obtains module 810 and includes:Initial value obtains unit, for according to target credit
Degree obtains the corresponding initial resource transfer numerical value of active user.Behavior pattern determination unit is tied for being compared according to abnormal position
Fruit determines the behavior pattern of active user.Target value obtains unit, for shifting number according to behavior pattern and initial resource
It is worth to the corresponding target resource transfer numerical value of active user.
Specific about information processing unit limits the restriction that may refer to above for information processing method, herein not
It repeats again.Modules in above- mentioned information processing unit can be realized fully or partially through software, hardware and combinations thereof.On
Stating each module can be embedded in or independently of in the processor in computer equipment, can also store in a software form in the form of hardware
In memory in computer equipment, the corresponding operation of the above modules is executed in order to which processor calls.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in figure 11.The computer equipment includes processor, memory and the network interface connected by system bus.
Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory of the computer equipment includes non-easy
The property lost storage medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program.The built-in storage
Operation for operating system and computer program in non-volatile memory medium provides environment.The network of the computer equipment connects
Mouth with external terminal by network connection for being communicated.To realize at a kind of information when the computer program is executed by processor
Reason method.
It will be understood by those skilled in the art that structure shown in Figure 11, only with the relevant part of application scheme
The block diagram of structure, does not constitute the restriction for the computer equipment being applied thereon to application scheme, and specific computer is set
Standby may include either combining certain components than more or fewer components as shown in the figure or being arranged with different components.
In one embodiment, a kind of computer equipment is provided, including memory, processor and storage on a memory and can located
The computer program run on reason device, processor realize following steps when executing computer program:Active user is obtained first
The multiple second position data of multiple first position data of period and active user in second time period;Obtain each
The corresponding consumption data of one position data obtains corresponding consumption feature according to each consumption data;Consumption feature input is pre-
In the credit rating model that first training obtains, the corresponding target credit rating of active user is obtained;By second position data and default position
It sets data to be compared, obtains abnormal position comparison result;Worked as according to abnormal position comparison result and target credit rating
The corresponding resource transfers strategy of preceding user.
In one embodiment, consumption feature is inputted in the credit rating model that training obtains in advance, obtains active user
Target credit rating the step of include:Consumption feature is separately input in the credit rating model that multiple advance training obtain, is obtained
The initial user's credit of the active user exported to each credit rating model;According to each initial user's credit and corresponding credit rating
The weight calculation of model obtains the corresponding target credit rating of active user.
In one embodiment, following steps are also realized when processor executes computer program:It obtains and carries out model training
Sample set, sample set includes multiple samples, and sample includes multiple trained consumption features and corresponding sample credit rating;According to
Sample set and a variety of different model training methods carry out model training, and it is trained to obtain each different model training method
The multiple credit rating models arrived;The corresponding consumption feature of sample is input in credit rating model, the corresponding model of sample is obtained
Credit rating;Each credit is obtained according to the gap of the corresponding model credit rating of sample in each credit rating model and sample credit rating
Spend the corresponding weight of model.
In one embodiment, performed by processor:According to the corresponding model credit rating of sample in each model and sample
The gap of this credit rating obtains the step of each model corresponding weight and includes:It is corresponding to calculate sample in each credit rating model
The deviation of model credit rating and sample credit rating;Read group total is carried out to the corresponding deviation of each credit rating model, is obtained each
The corresponding total deviation of credit rating model;It is obtained respectively according to the corresponding total deviation of each credit rating model and preset Weight algorithm
The corresponding weight of a credit rating model, wherein total deviation and weight are negative correlativing relation in Weight algorithm.
In one embodiment, the corresponding consumption data of each first position data of acquisition performed by processor, according to
The step of each consumption data obtains corresponding consumption feature include:The corresponding merchant information of each first position data is obtained,
Merchant information includes the consumption data of trade company and the merchant type of trade company;According to the consumption data of each trade company and trade company
Consumption type obtains corresponding consumption feature.
In one embodiment, predeterminated position data include active user in the historical actions track of third period, and
Two position datas include active user in the goal activity track of second time period, performed by processor by second position data
The step of being compared with predeterminated position data, obtaining abnormal position comparison result include:Active user is obtained in the third time
The historical actions track of section, the time where the third period is earlier than the time where second time period;By active user
The goal activity track of two periods is compared with active user in the historical actions track of third period, obtains exception bits
Set comparison result.
In one embodiment, being worked as according to abnormal position comparison result and target credit rating performed by processor
The step of the corresponding resource transfers strategy of preceding user includes:The corresponding initial resource of active user is obtained according to target credit rating to turn
Move numerical value;The behavior pattern of active user is determined according to abnormal position comparison result;Turned according to behavior pattern and initial resource
It moves numerical value and obtains the corresponding target resource transfer numerical value of active user.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program realizes following steps when being executed by processor:Obtain active user first time period multiple first position data with
And active user is in multiple second position data of second time period;The corresponding consumption data of each first position data is obtained,
Corresponding consumption feature is obtained according to each consumption data;Consumption feature is inputted in the credit rating model that training obtains in advance,
Obtain the corresponding target credit rating of active user;Second position data and predeterminated position data are compared, exception bits are obtained
Set comparison result;The corresponding resource transfers strategy of active user is obtained according to abnormal position comparison result and target credit rating.
In one embodiment, consumption feature is inputted in the credit rating model that training obtains in advance, obtains active user
Target credit rating the step of include:Consumption feature is separately input in the credit rating model that multiple advance training obtain, is obtained
The initial user's credit of the active user exported to each credit rating model;According to each initial user's credit and corresponding credit rating
The weight calculation of model obtains the corresponding target credit rating of active user.
In one embodiment, following steps are also realized when processor executes computer program:It obtains and carries out model training
Sample set, sample set includes multiple samples, and sample includes multiple trained consumption features and corresponding sample credit rating;According to
Sample set and a variety of different model training methods carry out model training, and it is trained to obtain each different model training method
The multiple credit rating models arrived;The corresponding consumption feature of sample is input in credit rating model, the corresponding model of sample is obtained
Credit rating;Each credit is obtained according to the gap of the corresponding model credit rating of sample in each credit rating model and sample credit rating
Spend the corresponding weight of model.
In one embodiment, performed by processor according to the corresponding model credit rating of sample in each model and sample
The gap of credit rating obtains the step of each model corresponding weight and includes:Calculate the corresponding mould of sample in each credit rating model
The deviation of type credit rating and sample credit rating;Read group total is carried out to the corresponding deviation of each credit rating model, obtains each letter
The corresponding total deviation of expenditure model;It is obtained according to the corresponding total deviation of each credit rating model and preset Weight algorithm each
The corresponding weight of credit rating model, wherein total deviation and weight are negative correlativing relation in Weight algorithm.
In one embodiment, the corresponding consumption data of each first position data of acquisition performed by processor, according to
The step of each consumption data obtains corresponding consumption feature include:The corresponding merchant information of each first position data is obtained,
Merchant information includes the consumption data of trade company and the merchant type of trade company;According to the consumption data of each trade company and trade company
Consumption type obtains corresponding consumption feature.
In one embodiment, predeterminated position data include active user in the historical actions track of third period, and
Two position datas include active user in the goal activity track of second time period, performed by processor by second position data
The step of being compared with predeterminated position data, obtaining abnormal position comparison result include:Active user is obtained in the third time
The historical actions track of section, the time where the third period is earlier than the time where second time period;By active user
The goal activity track of two periods is compared with active user in the historical actions track of third period, obtains exception bits
Set comparison result.
In one embodiment, being worked as according to abnormal position comparison result and target credit rating performed by processor
The step of the corresponding resource transfers strategy of preceding user includes:The corresponding initial resource of active user is obtained according to target credit rating to turn
Move numerical value;The behavior pattern of active user is determined according to abnormal position comparison result;Turned according to behavior pattern and initial resource
It moves numerical value and obtains the corresponding target resource transfer numerical value of active user.
One of ordinary skill in the art will appreciate that realizing all or part of flow in above-described embodiment method, being can be with
Instruct relevant hardware to complete by computer program, computer program can be stored in a non-volatile computer readable
It takes in storage medium, the computer program is when being executed, it may include such as the flow of the embodiment of above-mentioned each method.Wherein, this Shen
Any reference to memory, storage, database or other media used in each embodiment please provided, may each comprise
Non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above example can be combined arbitrarily, to keep description succinct, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield is all considered to be the range of this specification record.
Above example only expresses the several embodiments of the application, the description thereof is more specific and detailed, but can not
Therefore it is construed as limiting the scope of the patent.It should be pointed out that for those of ordinary skill in the art,
Under the premise of not departing from the application design, various modifications and improvements can be made, these belong to the protection domain of the application.
Therefore, the protection domain of the application patent should be determined by the appended claims.
Claims (10)
1. a kind of information processing method, the method includes:
Active user is obtained in multiple first position data of first time period and the active user in second time period
Multiple second position data;
The corresponding consumption data of each first position data is obtained, corresponding consumption is obtained according to each consumption data
Feature;
The consumption feature is inputted in the credit rating model that training obtains in advance, obtains the corresponding target letter of the active user
Expenditure;
The second position data are compared with predeterminated position data, obtain abnormal position comparison result;
The corresponding resource transfers of the active user are obtained according to the abnormal position comparison result and the target credit rating
Strategy.
2. according to the method described in claim 1, it is characterized in that, described input what training in advance obtained by the consumption feature
In credit rating model, the step of obtaining the target credit rating of the active user, includes:
The consumption feature is separately input in the credit rating model that multiple advance training obtain, obtains each letter
The initial user's credit of the active user of expenditure model output;
The active user is obtained according to the weight calculation of each initial user's credit and corresponding credit rating model to correspond to
Target credit rating.
3. according to the method described in claim 2, it is characterized in that, the method further includes:
The sample set for carrying out model training is obtained, the sample set includes multiple samples, and the sample includes multiple training consumption
Feature and corresponding sample credit rating;
Model training is carried out according to the sample set and a variety of different model training methods, obtains each different mould
Multiple credit rating models that type training method is trained;
The corresponding consumption feature of the sample is input in the credit rating model, the corresponding model credit of the sample is obtained
Degree;
Institute is obtained according to the gap of the corresponding model credit rating of sample in each credit rating model and the sample credit rating
State the corresponding weight of each credit rating model.
4. according to the method described in claim 3, it is characterized in that, described according to the corresponding model of sample in each model
Credit rating and the gap of the sample credit rating obtain the step of each model corresponding weight and include:
Calculate the deviation of sample corresponding model credit rating and the sample credit rating in each credit rating model;
Read group total is carried out to the corresponding deviation of each credit rating model, it is corresponding total to obtain each credit rating model
Deviation;
Each credit rating mould is obtained according to each corresponding total deviation of credit rating model and preset Weight algorithm
The corresponding weight of type, wherein total deviation and weight are negative correlativing relation in the Weight algorithm.
5. according to the method described in claim 1, it is characterized in that, described obtain that each first position data are corresponding to disappear
The step of taking data, corresponding consumption feature is obtained according to each consumption data include:
The corresponding merchant information of each first position data is obtained, the merchant information includes the consumption data of the trade company
And the merchant type of the trade company;
Corresponding consumption feature is obtained according to the consumption type of the consumption data of each trade company and the trade company.
6. according to the method described in claim 1, it is characterized in that, the predeterminated position data include the active user
The historical actions track of three periods, the second position data include goal activity of the active user in second time period
Track, the described the step of second position data are compared with predeterminated position data, obtain abnormal position comparison result
Including:
It is early in the time where the historical actions track of the third period, the third period to obtain the active user
Time where the second time period;
By the active user in the goal activity track of the second time period and the active user in the third time
The historical actions track of section is compared, and abnormal position comparison result is obtained.
7. according to the method described in claim 1, it is characterized in that, described according to the abnormal position comparison result and described
The step that target credit rating obtains the corresponding resource transfers strategy of the active user includes:
The corresponding initial resource of the active user, which is obtained, according to the target credit rating shifts numerical value;
The behavior pattern of the active user is determined according to the abnormal position comparison result;
The corresponding target resource of the active user is obtained according to the behavior pattern and initial resource transfer numerical value to turn
Move numerical value.
8. a kind of information processing unit, described device include:
Position data acquisition module, for obtaining active user in multiple first position data of first time period and described working as
Multiple second position data of the preceding user in second time period;
Feature obtains module, for obtaining the corresponding consumption data of each first position data, according to each consumption
Data obtain corresponding consumption feature;
Credit rating obtains module, for inputting the consumption feature in the credit rating model that training obtains in advance, obtains described
The corresponding target credit rating of active user;
Contrast module obtains abnormal position and compares knot for comparing the second position data with predeterminated position data
Fruit;
Strategy obtains module, for obtaining the current use according to the abnormal position comparison result and the target credit rating
The corresponding resource transfers strategy in family.
9. a kind of computer equipment, including memory, processor and storage are on a memory and the meter that can run on a processor
Calculation machine program, which is characterized in that the processor realizes any one of claim 1 to 7 institute when executing the computer program
The step of stating method.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claim 1 to 7 is realized when being executed by processor.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810102329.7A CN108460674A (en) | 2018-02-01 | 2018-02-01 | Information processing method, device, computer equipment and storage medium |
PCT/CN2018/088985 WO2019148715A1 (en) | 2018-02-01 | 2018-05-30 | Information processing method and apparatus, and computer device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810102329.7A CN108460674A (en) | 2018-02-01 | 2018-02-01 | Information processing method, device, computer equipment and storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108460674A true CN108460674A (en) | 2018-08-28 |
Family
ID=63238550
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810102329.7A Pending CN108460674A (en) | 2018-02-01 | 2018-02-01 | Information processing method, device, computer equipment and storage medium |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN108460674A (en) |
WO (1) | WO2019148715A1 (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110060144A (en) * | 2019-03-18 | 2019-07-26 | 平安科技(深圳)有限公司 | Amount model training method, amount appraisal procedure, device, equipment and medium |
CN110751395A (en) * | 2019-10-21 | 2020-02-04 | 中国民航信息网络股份有限公司 | Passenger travel state determination method, device and server |
CN111860647A (en) * | 2020-07-21 | 2020-10-30 | 金陵科技学院 | Abnormal consumption mode judgment method |
CN113724426A (en) * | 2021-06-24 | 2021-11-30 | 储信(北京)科技发展有限公司 | Multi-dimensional security rule configuration early warning method |
CN114444682A (en) * | 2021-10-18 | 2022-05-06 | 中国联合网络通信集团有限公司 | Training method and device of neural network model |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110795467B (en) * | 2019-09-26 | 2024-02-27 | 腾讯大地通途(北京)科技有限公司 | Traffic rule data processing method, device, storage medium and computer equipment |
CN111523765B (en) * | 2020-03-25 | 2024-03-22 | 平安科技(深圳)有限公司 | Material demand analysis method, equipment, device and readable storage medium |
CN112017060B (en) * | 2020-07-15 | 2024-04-26 | 北京淇瑀信息科技有限公司 | Method and device for allocating resources for target user and electronic equipment |
CN115203292B (en) * | 2022-09-15 | 2022-11-25 | 昆仑智汇数据科技(北京)有限公司 | Data processing method, device and equipment for industrial equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104866969A (en) * | 2015-05-25 | 2015-08-26 | 百度在线网络技术(北京)有限公司 | Personal credit data processing method and device |
CN105894089A (en) * | 2016-04-21 | 2016-08-24 | 百度在线网络技术(北京)有限公司 | Method of establishing credit investigation model, credit investigation determination method and the corresponding apparatus thereof |
CN106447489A (en) * | 2016-09-12 | 2017-02-22 | 中山大学 | Partially stacking blend based user credit assessment model |
CN106530078A (en) * | 2016-11-29 | 2017-03-22 | 流量海科技成都有限公司 | Loan risk early warning method and system based on multi-industry data |
CN106934602A (en) * | 2017-03-12 | 2017-07-07 | 上海云剑信息技术有限公司 | Based on the personal finance credit assessment method that geographical, motion is merged with social information |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105809408A (en) * | 2014-12-30 | 2016-07-27 | 金蝶软件(中国)有限公司 | Data transmission method and data transmission device |
CN107464180A (en) * | 2016-06-02 | 2017-12-12 | 张岩 | A kind of student's credit-graded approach of Behavior-based control data |
CN107220236A (en) * | 2017-05-23 | 2017-09-29 | 武汉朱雀闻天科技有限公司 | It is a kind of to determine the doubtful naked method and device for borrowing student |
-
2018
- 2018-02-01 CN CN201810102329.7A patent/CN108460674A/en active Pending
- 2018-05-30 WO PCT/CN2018/088985 patent/WO2019148715A1/en active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104866969A (en) * | 2015-05-25 | 2015-08-26 | 百度在线网络技术(北京)有限公司 | Personal credit data processing method and device |
CN105894089A (en) * | 2016-04-21 | 2016-08-24 | 百度在线网络技术(北京)有限公司 | Method of establishing credit investigation model, credit investigation determination method and the corresponding apparatus thereof |
CN106447489A (en) * | 2016-09-12 | 2017-02-22 | 中山大学 | Partially stacking blend based user credit assessment model |
CN106530078A (en) * | 2016-11-29 | 2017-03-22 | 流量海科技成都有限公司 | Loan risk early warning method and system based on multi-industry data |
CN106934602A (en) * | 2017-03-12 | 2017-07-07 | 上海云剑信息技术有限公司 | Based on the personal finance credit assessment method that geographical, motion is merged with social information |
Non-Patent Citations (2)
Title |
---|
卢少平 等: "《物流信息技术与应用》", 30 September 2009, 华中科技大学出版社 * |
欧冬秀: "《交通信息技术 第2版》", 31 March 2014, 同济大学出版社 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110060144A (en) * | 2019-03-18 | 2019-07-26 | 平安科技(深圳)有限公司 | Amount model training method, amount appraisal procedure, device, equipment and medium |
CN110060144B (en) * | 2019-03-18 | 2024-01-30 | 平安科技(深圳)有限公司 | Method for training credit model, method, device, equipment and medium for evaluating credit |
CN110751395A (en) * | 2019-10-21 | 2020-02-04 | 中国民航信息网络股份有限公司 | Passenger travel state determination method, device and server |
CN110751395B (en) * | 2019-10-21 | 2023-07-04 | 中国民航信息网络股份有限公司 | Passenger journey state determining method, device and server |
CN111860647A (en) * | 2020-07-21 | 2020-10-30 | 金陵科技学院 | Abnormal consumption mode judgment method |
CN111860647B (en) * | 2020-07-21 | 2023-11-10 | 金陵科技学院 | Abnormal consumption mode judging method |
CN113724426A (en) * | 2021-06-24 | 2021-11-30 | 储信(北京)科技发展有限公司 | Multi-dimensional security rule configuration early warning method |
CN114444682A (en) * | 2021-10-18 | 2022-05-06 | 中国联合网络通信集团有限公司 | Training method and device of neural network model |
CN114444682B (en) * | 2021-10-18 | 2023-01-24 | 中国联合网络通信集团有限公司 | Training method and device of neural network model |
Also Published As
Publication number | Publication date |
---|---|
WO2019148715A1 (en) | 2019-08-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108460674A (en) | Information processing method, device, computer equipment and storage medium | |
Balboni | In harm's way? infrastructure investments and the persistence of coastal cities | |
CN108446281B (en) | Method, device and storage medium for determining user intimacy | |
Zhong et al. | Inferring building functions from a probabilistic model using public transportation data | |
CN105532030B (en) | For analyzing the devices, systems, and methods of the movement of target entity | |
AU2013364041B2 (en) | Instance weighted learning machine learning model | |
de Oliveira et al. | Evaluating, partitioning, and mapping the spatial autocorrelation component in ecological niche modeling: a new approach based on environmentally equidistant records | |
CN109657837A (en) | Default Probability prediction technique, device, computer equipment and storage medium | |
CN110414732B (en) | Travel future trajectory prediction method and device, storage medium and electronic equipment | |
CN106097095B (en) | Determine the method and device of credit | |
Frias-Martinez et al. | Towards large scale technology impact analyses: Automatic residential localization from mobile phone-call data | |
CN109767319A (en) | The accrediting amount determines method, apparatus, computer equipment and storage medium | |
CN107633100A (en) | A kind of point of interest based on incorporation model recommends method and device | |
CN110069706A (en) | Method, end side equipment, cloud side apparatus and the end cloud cooperative system of data processing | |
CN108932582A (en) | Risk information determines method, apparatus, computer equipment and storage medium | |
CN109769210A (en) | User Activity Regional Similarity judgment method, device, computer equipment | |
CN109584045A (en) | Accrediting amount configuration method, device, computer equipment and storage medium | |
CN110147923B (en) | Method and device for identifying risk users | |
CN109949063A (en) | A kind of address determines method, apparatus, electronic equipment and readable storage medium storing program for executing | |
CN109947870A (en) | The prediction meanss and method of specific type personnel calculate equipment and storage medium | |
CN108182633A (en) | Loan data processing method, device, computer equipment and storage medium | |
CN110348896A (en) | Divide the method for geographic grid, commercial circle determines method and apparatus | |
CN113449986A (en) | Service distribution method, device, server and storage medium | |
CN108932525A (en) | A kind of behavior prediction method and device | |
CN108304482A (en) | The recognition methods and device of broker, electronic equipment and readable storage medium storing program for executing |
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: 20180828 |