CN110276668A - The method and system that finance product intelligently pushing, matching degree determine - Google Patents
The method and system that finance product intelligently pushing, matching degree determine Download PDFInfo
- Publication number
- CN110276668A CN110276668A CN201910583510.9A CN201910583510A CN110276668A CN 110276668 A CN110276668 A CN 110276668A CN 201910583510 A CN201910583510 A CN 201910583510A CN 110276668 A CN110276668 A CN 110276668A
- Authority
- CN
- China
- Prior art keywords
- information
- client
- training
- finance product
- customer
- 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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
-
- 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
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0639—Item locations
-
- 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
- G06Q40/06—Asset management; Financial planning or analysis
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Human Resources & Organizations (AREA)
- Operations Research (AREA)
- Technology Law (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
It is provided by the invention provide a kind of finance product intelligently pushing, the method and system that matching degree determines, introduce the machine learning modeling method of artificial intelligence field, more accurate and intelligent financing scheme can be exported with curstomer-oriented through the invention, meet the needs of individualities of enterprises, effectively promotes the matching degree of finance product and the usage experience of client.
Description
Technical field
The present invention relates to legal person's financing marketing and artificial intelligence fields, more particularly to finance product intelligently pushing, matching
Spend determining method and system.
Background technique
Finance product be business bank according to different business entitys, mechanism legal person etc. to public client or other to private visitor
Family demand and the finance product for designing sale are the important sources taken among business bank's service for corporate customers.Finance product at present
Marketing program mainly handles artificial formulation by bank client, and there are the following problems for this mode: 1, due to customer information
Limited with product information grasp, practical marketing program can not really agree with customer demand, play " thousand people, thousand face " precision marketing
Effect.2, customer manager manually formulates marketing program often iteration cycle is longer, can not cope with flexible and changeable market.3, client
It handles ordinary priority follow-up and safeguards large and medium-sized state-owned enterprise, individual enterprise client, wider medium-sized and small enterprises are covered incomplete.
Not yet occur for the relevant technologies of the intelligent recommendation of finance product at present, the degree of automation is low, the above problem one
It can not directly solve.
Summary of the invention
In order to solve the problems, such as that current finance product formulates at least one of caused, the application by bank client manager
There is provided a kind of finance product intelligently pushing, the method and system that matching degree determines.
In the embodiment of the one aspect of the application, a kind of finance product intelligently pushing method is provided, comprising:
Obtain the identity of client to be pushed;
By client's archive information of the correspondence identity and the preset prediction model of customer action information input, obtain
The push of finance product is as a result, the prediction model is obtained by forecast sample training;
Push the push result;
Wherein, the forecast sample includes the reason of client's archive information, customer action information, the client selection of multiple clients
Wealth product information and current bank marketing message.
In certain embodiments, client's archive information includes: that client opens an account information, customer capital information and client
Credit information.
In certain embodiments, the customer action information includes: financing behavioural information and trading activity information.
In certain embodiments, the finance product information includes: product risks rank, issuing way and prospective earnings
Rate.
In certain embodiments, further includes:
Establish initial predicted model;
Pass through the forecast sample training initial predicted model.
In certain embodiments, the client's archive information and customer action information input by the correspondence identity
Preset prediction model, comprising:
Characteristic processing is carried out the client's archive information and customer action information of the correspondence identity, is generated corresponding
Feature;
The feature is input to the prediction model, wherein the forecast sample is by client's archive information, Ke Huhang
The feature set formed after the finance product information and current bank marketing message characteristic processing that are selected for information, client.
In certain embodiments, the step of characteristic processing of the forecast sample includes:
The up-to-date information set in duration is differed with current time from screening in the information to characteristic processing;
The up-to-date information is screened, removal obtains foundation characteristic after showing unconverted information to all clients;
The corresponding foundation characteristic of the customer action information is analyzed, customer action temporal aspect is obtained;
The corresponding foundation characteristic of the finance product information is analyzed, product variations temporal aspect is obtained;
According to the customer action temporal aspect and the product variations temporal aspect, customer portrait feature is constructed, it is described
Customer portrait feature includes: client's liveness, client's channel preference, client's financing risk partiality, clients fund retention ratio, clearing
Concentration degree of trading and client's financing risk coefficient.
In certain embodiments, the step of characteristic processing of the forecast sample further include:
Reject the feature that the degree of correlation in the customer portrait feature is lower than given threshold;
Feature Dimension Reduction processing is carried out to the customer portrait feature after rejecting.
It is in certain embodiments, described to pass through the forecast sample training initial predicted model, comprising:
Called data processing rule and model training rule;
Data processing is carried out to the forecast sample by the data processing rule;
To data, treated that the forecast sample carries out characteristic processing, obtains characteristic set;
By the feature in conjunction with the initial predicted model is input to, according to the model training rule training initial predicted
Model;
Grey iterative generation model training threshold value updates institute based on the size relation of the model training threshold value and preset threshold
Forecast sample is stated, replaces former forecast sample with the forecast sample of update, until model training threshold value is greater than preset threshold, into
And obtain the prediction model of training completion.
In certain embodiments, the data processing rule includes:
Dealing of abnormal data, amount of money Logarithm conversion, the conversion of date integer;
The model training rule includes: training algorithm, positive and negative sample proportion, training set and verifying collection ratio, training frequency
Rate.
In further aspect of the application embodiment, the matching degree of a kind of client and finance product determines method, comprising:
Obtain the client's archive information, customer action information, the product information of finance product to be matched of client to be matched with
And current bank marketing message;
By client's archive information, customer action information, the product information of finance product to be matched and current bank
Marketing message inputs preset prediction model, generates the matching degree of client to be matched Yu the finance product to be matched;
Wherein, the prediction model is obtained by forecast sample training, and the forecast sample includes the visitor of multiple clients
Family archive information, customer action information, the finance product information and current bank marketing message of client's selection.
In certain embodiments, client's archive information includes: that client opens an account information, customer capital information and client
Credit information.
In certain embodiments, the customer action information includes: financing behavioural information and trading activity information.
In certain embodiments, the finance product information includes: product risks rank, issuing way and prospective earnings
Rate.
In certain embodiments, further includes:
Establish initial predicted model;
Pass through the forecast sample training initial predicted model.
In certain embodiments, the client's archive information and customer action information input by the correspondence identity
Preset prediction model, comprising:
Characteristic processing is carried out the client's archive information and customer action information of the correspondence identity, is generated corresponding
Feature;
The feature is input to the prediction model, wherein the forecast sample is by client's archive information, Ke Huhang
The feature set formed after the finance product information and current bank marketing message characteristic processing that are selected for information, client.
In certain embodiments, the step of characteristic processing of the forecast sample includes:
The up-to-date information set in duration is differed with current time from screening in the information to characteristic processing;
The up-to-date information is screened, removal obtains foundation characteristic after showing unconverted information to all clients;
The corresponding foundation characteristic of the customer action information is analyzed, customer action temporal aspect is obtained;
The corresponding foundation characteristic of the finance product information is analyzed, product variations temporal aspect is obtained;
According to the customer action temporal aspect and the product variations temporal aspect, customer portrait feature is constructed, it is described
Customer portrait feature includes: client's liveness, client's channel preference, client's financing risk partiality, clients fund retention ratio, clearing
Concentration degree of trading and client's financing risk coefficient.
In certain embodiments, the step of characteristic processing of the forecast sample further include:
Reject the feature that the degree of correlation in the customer portrait feature is lower than given threshold;
Feature Dimension Reduction processing is carried out to the customer portrait feature after rejecting.
It is in certain embodiments, described to pass through the forecast sample training initial predicted model, comprising:
Called data processing rule and model training rule;
Data processing is carried out to the forecast sample by the data processing rule;
To data, treated that the forecast sample carries out characteristic processing, obtains characteristic set;
By the feature in conjunction with the initial predicted model is input to, according to the model training rule training initial predicted
Model;
Grey iterative generation model training threshold value updates institute based on the size relation of the model training threshold value and preset threshold
Forecast sample is stated, replaces former forecast sample with the forecast sample of update, until model training threshold value is greater than preset threshold, into
And obtain the prediction model of training completion.
In certain embodiments, the data processing rule includes:
Dealing of abnormal data, amount of money Logarithm conversion, the conversion of date integer;
The model training rule includes: training algorithm, positive and negative sample proportion, training set and verifying collection ratio, training frequency
Rate.
In the another aspect embodiment of the application, a kind of finance product intelligently pushing system, comprising:
First obtains module, obtains the identity of client to be pushed;
Result-generation module is pushed, client's archive information of the correspondence identity and customer action information input is pre-
If prediction model, obtain the push of finance product as a result, the prediction model be by forecast sample training obtain;
Pushing module pushes the push result;
Wherein, the forecast sample includes the reason of client's archive information, customer action information, the client selection of multiple clients
Wealth product information and current bank marketing message.
In the another aspect embodiment of the application, the matching degree of a kind of client and finance product determines system, comprising:
Second obtains module, obtains the client's archive information, customer action information, finance product to be matched of client to be matched
Product information and current bank marketing message;
Matching module, by client's archive information, customer action information, the product information of finance product to be matched and
Current bank marketing message inputs preset prediction model, generates the matching of client to be matched Yu the finance product to be matched
Degree;
Wherein, the prediction model is obtained by forecast sample training, and the forecast sample includes the visitor of multiple clients
Family archive information, customer action information, the finance product information and current bank marketing message of client's selection.
In the another aspect embodiment of the application, a kind of computer equipment, including memory, processor and be stored in
On reservoir and the computer program that can run on a processor, the processor realize side as described above when executing described program
The step of method.
In the another aspect embodiment of the application, a kind of computer readable storage medium is stored thereon with computer journey
The step of sequence, which realizes method as described above when being executed by processor.
The invention has the following beneficial effects:
It is provided by the invention to provide the determining method and system of finance product intelligently pushing, matching degree, introduce artificial intelligence
The machine learning modeling method in energy field, can export more accurate and intelligent financing scheme through the invention with curstomer-oriented,
Meet the needs of individualities of enterprises, effectively promotes the matching degree of finance product and the usage experience of client.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the system relationship schematic diagram of legal person's financing intelligent recommendation system in the embodiment of the present invention.
Fig. 2 is the structural block diagram of legal person's financing intelligent recommendation system in the embodiment of the present invention.
Fig. 3 is data acquisition device function structure chart in the embodiment of the present invention.
Fig. 4 is model training apparatus function structure chart in the embodiment of the present invention.
Fig. 5 is prediction meanss function structure chart in the embodiment of the present invention.
Fig. 6 is marketing apparatus function structure chart in the embodiment of the present invention.
Fig. 7 is data acquisition module structural schematic diagram in the embodiment of the present invention.
Fig. 8 is data memory module structural schematic diagram in the embodiment of the present invention.
Fig. 9 is feature processing block structural schematic diagram in the embodiment of the present invention.
Figure 10 is the working-flow figure of legal person's financing intelligent recommendation system in the embodiment of the present invention.
Figure 11 is characteristic processing flow chart in the embodiment of the present invention.
Figure 12 is a kind of idiographic flow schematic diagram of finance product intelligently pushing method in the embodiment of the present invention.
Figure 13 is the establishment step schematic diagram of prediction model in the embodiment of the present invention.
Figure 14 is the idiographic flow schematic diagram of step S2a in Figure 12 in the embodiment of the present invention.
Figure 15 is that the matching degree of a kind of client and finance product determines method flow schematic diagram in the embodiment of the present invention.
Figure 16 is a kind of virtual module structural schematic diagram of finance product intelligently pushing system in the embodiment of the present invention.
Figure 17 is the virtual module structure that the matching degree of a kind of client and finance product determines system in the embodiment of the present invention
Schematic diagram.
Figure 18 shows the structural schematic diagram for being suitable for the computer equipment for being used to realize the embodiment of the present application.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
With the maturation of the technologies such as mobile Internet, big data, cloud computing, it is new that artificial intelligence has become global scientific and technological giant
Strategic development direction, more and more industries start to embrace artificial intelligence, are developed with the upgrading of " artificial intelligence+" power-assisted industry.
How by big data and artificial intelligence technology involvement client marketing and service process, business bank's big data information valence is given full play to
Value, towards corporate client establish completely new low cost, high efficiency financing sales mode, meet enterprise it is more personalized and embody
Demand is extremely urgent.
Figure 12 shows a kind of flow diagram of finance product intelligently pushing method in the embodiment of the present application, such as Figure 12 institute
Show, the finance product intelligently pushing method, comprising:
S1a: the identity of client to be pushed is obtained;
S2a: by client's archive information of the correspondence identity and the preset prediction model of customer action information input,
The push of finance product is obtained as a result, the prediction model is obtained by forecast sample training;
S3a: the push result is pushed.
The identity of client refers to the flag information for being determined for client identity, such as the visitor inside banking system
Family coding, customer's identity card etc. can be used as the flag information of client identity.
Forecast sample in the application, client's archive information, customer action information, client's selection including multiple clients
Finance product information and current bank marketing message.The system typing above- mentioned information for passing through bank itself, use above- mentioned information
Prediction model is trained, so that stable prediction model is obtained, since the prediction model is using the training of a large amount of historical datas
It is formed, therefore intelligence degree and order of accuarcy are higher, while prediction model is obtained by training, in actual use often
The partial data of one client is equivalent to a forecast sample, and then constantly improve and update prediction model, so as to height
The finance product that effect is steadily suitble to for lead referral, provides the usage experience of client.
Client generally comprises two classes, to public client and individual client, business entity etc. is generally to public client, due to public affairs
The fund scale of construction of client and individual client, demand etc. are all different, such as the stability of fund is more lain in public client, and a
People client may partly belong to radical type investor, partly belong to robustness investor, and then the otherness of individual client is larger,
In view of above-mentioned otherness, two independent prediction models are generally used to public client and individual client, the two is not interfere with each other.
The application is applicable to be equally applicable to individual client to public client, and difference is specific prediction model
Configuration it is different, this will not be repeated here.
Generally, either to public client or individual client, client's archive information include: client open an account information,
Customer capital information and credit information.
It is appreciated that the client in the embodiment of the present application can be new client (the not client of customer data in system),
It is also possible to frequent customer's (data are stored in the client in system) to need as trained customers with historical data certainly,
Namely frequent customer, this will not be repeated here.
In some embodiments, the customer action information includes: financing behavioural information and trading activity information.
Further, in some embodiments, the finance product information include: product risks rank, issuing way and
Expected yield.
The application above method further includes the steps that establishing and improves prediction model, and as shown in figure 13, i.e., the above method is also
Include:
S001: initial predicted model is established;
S002: pass through the forecast sample training initial predicted model.
Figure 14 shows the idiographic flow schematic diagram of step S200 in an embodiment.Step S2 is specifically included:
S201: characteristic processing is carried out the client's archive information and customer action information of the correspondence identity, is generated
Corresponding feature;
S202: being input to the prediction model for the feature, wherein the forecast sample is by client's archive information, visitor
The feature set formed after family behavioural information, the finance product information and current bank marketing message characteristic processing of client's selection.
In the embodiment, by above-mentioned mixed and disorderly respective information characteristics, and then facilitates and compare and handle.
Further, the step of characteristic processing of the forecast sample includes:
The up-to-date information set in duration is differed with current time from screening in the information to characteristic processing;
The up-to-date information is screened, removal obtains foundation characteristic after showing unconverted information to all clients;
The corresponding foundation characteristic of the customer action information is analyzed, customer action temporal aspect is obtained;
The corresponding foundation characteristic of the finance product information is analyzed, product variations temporal aspect is obtained;
According to the customer action temporal aspect and the product variations temporal aspect, customer portrait feature is constructed, it is described
Customer portrait feature includes: client's liveness, client's channel preference, client's financing risk partiality, clients fund retention ratio, clearing
Concentration degree of trading and client's financing risk coefficient.
Further, the step of characteristic processing of the forecast sample further include:
Reject the feature that the degree of correlation in the customer portrait feature is lower than given threshold;
Feature Dimension Reduction processing is carried out to the customer portrait feature after rejecting.
It is described to pass through the forecast sample training initial predicted model in some embodiments, comprising:
Called data processing rule and model training rule;
Data processing is carried out to the forecast sample by the data processing rule;
To data, treated that the forecast sample carries out characteristic processing, obtains characteristic set;
By the feature in conjunction with the initial predicted model is input to, according to the model training rule training initial predicted
Model;
Grey iterative generation model training threshold value updates institute based on the size relation of the model training threshold value and preset threshold
Forecast sample is stated, replaces former forecast sample with the forecast sample of update, until model training threshold value is greater than preset threshold, into
And obtain the prediction model of training completion.
In the embodiment, the data processing rule includes:
Dealing of abnormal data, amount of money Logarithm conversion, the conversion of date integer;
The model training rule includes: training algorithm, positive and negative sample proportion, training set and verifying collection ratio, training frequency
Rate.
The push that can be not only used for finance product using the prediction model of the application, can be used for known client with
Matching degree between specific finance product calculates, that is, knows that certain a finance product is appropriate for the client.
Figure 15 shows the matching degree determination side of a kind of client and finance product of the application another aspect embodiment offer
Method, comprising:
S1b: obtain client's archive information of client to be matched, customer action information, finance product to be matched product letter
Breath and current bank marketing message;
S2b: by client's archive information, customer action information, the product information of finance product to be matched and current
The preset prediction model of marketing of bank information input generates the matching degree of client to be matched Yu the finance product to be matched;
Wherein, the prediction model is obtained by forecast sample training, and the forecast sample includes the visitor of multiple clients
Family archive information, customer action information, the finance product information and current bank marketing message of client's selection.
The matching degree of present aspect determines that method can automatically determine the matching degree between client and known finance product, in turn
So that customer self-service is seeked advice from its matching relationship between finance product, so as to independently select finance product, does not need to carry out
Attendant consultation service saves human and material resources, and improves the usage experience of user.
Based on identical reason, the specific example in above-mentioned finance product intelligently pushing method be can be used in present aspect
Matching degree determine in method, such as the step of establishing prediction model, training prediction model the step of and characteristic matching step
It is rapid etc., do not make exhaustion herein.
Based on identical reason, the another aspect embodiment of the application provides a kind of finance product intelligently pushing system, such as schemes
16 include: the first acquisition module 100a, obtains the identity of client to be pushed;Result-generation module 200a is pushed, will be corresponded to
The client's archive information and the preset prediction model of customer action information input of the identity, obtain the push of finance product
As a result, the prediction model is obtained by forecast sample training;Pushing module 300a pushes the push result;Wherein, institute
State forecast sample include multiple clients client's archive information, customer action information, client selection finance product information and
Current bank marketing message.
Based on identical reason, the another aspect embodiment of the application provides the matching degree determination of a kind of client and finance product
System, as shown in figure 17, comprising: second obtains module 100b, obtains client's archive information, the customer action letter of client to be matched
Breath, the product information of finance product to be matched and current bank marketing message;The client is achieved and is believed by matching module 200b
Breath, customer action information, the product information of finance product to be matched and current bank marketing message input preset prediction mould
Type generates the matching degree of client to be matched Yu the finance product to be matched;Wherein, the prediction model is to pass through forecast sample
Training obtains, and the forecast sample includes that the financing of client's archive information, customer action information, the client selection of multiple clients produces
Product information and current bank marketing message.
It is understood that the matching degree of finance product intelligently pushing system and client and finance product provided by the present application
It determines that the module in system can be device and constitute, can also be combined and be constituted by multiple devices, device can also be because
It can realize the function of multiple modules and the combination as multiple modules simultaneously, that is to say, that in a particular embodiment, due to upper
The module for stating system is according to function distinguishing, therefore device and module not correspond, and the application does not repeat them here.
Such as in following specific embodiments, the first acquisition module or the second acquisition module can be data acquisition dress
It sets, pushes result-generation module or matching module can be specially prediction meanss, pushing module can be specially marketing apparatus.
The above-mentioned finance product intelligently pushing method of the application is described in detail for public client below, still
It is to be understood that being only in that specific processing details is different for individual client, general plotting is identical as to public client.
Fig. 1 is the system relationship schematic diagram of legal person (to public client) financing intelligent recommendation system, legal person's finance product intelligence
Supplying system 1 is respectively with enterprise-level customer information system 2, to public client's basic service system 3, credit system 4, big data platform
5, legal person's marketing system 6, financing system 7, Web bank, enterprise 8, Mobile banking 9 connect, wherein legal person's finance product intelligently pushes away
Sending system 1 is core of the invention, enterprise-level customer information system 2, to public client's basic service system 3, credit system 4, big
Data platform 5, legal person's marketing system 6, financing system 7, Web bank, enterprise 8, Mobile banking 9 are existing out-out business system.
It is specific:
Legal person's finance product intelligently pushing system 1 is responsible for receiving the command information of existing business system and be located accordingly
Reason, described instruction information include but is not limited to that legal person's financing of legal person's marketing system 6, Web bank, enterprise 8, Mobile banking 9 pushes away
Recommend the model training instruction of instruction and big data platform 5, legal person's marketing system 6, legal person's finance product intelligently pushing system 1
Internal structure detailed description see Fig. 2.
Enterprise-level customer information system 2, to public client's basic service system 3, credit system 4, big data platform 5, legal person
Marketing system 6, financing system 7, Web bank, enterprise 8, Mobile banking 9 are existing business system, are responsible for providing legal person's financing production
Product intelligently pushing system data to be treated, and processing is received as a result, its concrete function repeats no more.
As shown in Fig. 2, a kind of legal person of the present invention manages money matters, intelligent recommendation system includes data acquisition device 11, model training dress
Set 12, prediction meanss 13, marketing apparatus 14, wherein data acquisition device 11 and model training apparatus 12, prediction meanss 13, battalion
Pin assembly 14 connects, and model training apparatus 12 is connect with data acquisition device 11, prediction meanss 13, and prediction meanss 13 are adopted with data
Acquisition means 11, model training apparatus 12, marketing apparatus 14 connect, and marketing apparatus 14 and data acquisition device 11, model training fill
The connection of 12, prediction meanss 13 is set, specific:
Data acquisition device 11 is responsible for receiving the data harvesting request of prediction meanss 13, model training apparatus 12, according to adopting
Collect data area and initiate data harvesting request to marketing apparatus 14, and acquisition data are stored, subsequent notification model training
Device 12 carries out data processing, and Fig. 3 is shown in detailed description.
Model training apparatus 12 is responsible for receiving the data processing request of data acquisition device 11, and carrys out self-timing batch
The request of the model training of system and marketing apparatus 14 updates data to data acquisition device 11 after the completion of processing, while notifying pre-
It surveys device 13 and carries out prediction processing, the timing batch system is the timing batch system of big data platform 5, is instructed respectively with model
Practice device 12, prediction meanss 13 be connected, by timing batch to legal person's finance product intelligently pushing system 1 initiate model training and
Fig. 4 is shown in predictions request, the detailed description of model training apparatus 12.
Prediction meanss 13, responsible reception carry out the predictions request of self-timing batch system and marketing apparatus 14, and notification data is adopted
Acquisition means 11 carry out data acquisition, and the instruction for then receiving model training apparatus 12 carries out prediction processing, and Fig. 5 is shown in detailed description.
Instruction and the mould from customer manager are recommended in marketing apparatus 14, the legal person's financing for being responsible for receiving out-out business system
Type train request then notifies prediction meanss 13 and model training apparatus 12 to carry out prediction processing, model training processing respectively, together
When, the data that marketing apparatus 14 is responsible for processing data acquisition device 11 acquire demand, from each on-demand gathering of out-out business system
Fig. 6 is shown in business data, detailed description.
Fig. 3 data acquisition device 11 includes data acquisition module 111 and data memory module 112, in which:
The responsible on-line prediction data harvesting request received from on-line prediction processing module 132 of data acquisition module 111,
Offline prediction data from offline prediction processing module 133 acquires request, and the number from training instruction processing module 121
It requests, collect business datum from peripheral system and is stored to data memory module 112, after the completion of storage at notification data according to acquisition
It manages module 122 and carries out data processing, internal structure chart is shown in Fig. 7.
Data memory module 112 is responsible for acquired data storage of the reception from data acquisition module 111 and instructs, comes from feature
The model training information store instruction and model prediction information store instruction of processing module 123, and come from model training mould
Data are respectively stored into corresponding memory block by the training pattern store instruction of block 124, and internal structure chart is shown in Fig. 8.
Fig. 4 model training apparatus 12 includes training instruction processing module 121, data processing module 122, feature processing block
123 and model training module 124, in which:
Training instruction processing module 121 is responsible for receiving the model of self-timing batch system and customer manager's marketing module 142
Train request and notification data acquisition module 111 complete collecting training data processing.
Data processing module 122 is responsible for receiving the data processing request from data acquisition module 111, is stored according to data
The data processing rule of 112 preset rules memory block of module pre-processes data.It is requested if it is training data processing, it will
Processing result is stored to the model training information storage area of data memory module 112;It requests, then will if it is prediction data processing
Processing result is stored to the model prediction information storage area of data memory module 112, and feature processing block 123 is notified to carry out spy
Levy working process.
Feature processing block 123 is responsible for receiving the characteristic processing request from data processing module 122, is stored from data
Module 112 obtains pending data and feature Expert Rules, and processing generates final characteristic set.If processing is trained number
According to feature, processing result is stored to the model training information storage area of data memory module 112, and notification model training module
124 carry out model training processing;If processing is prediction data feature, by processing result storage to data memory module 112
Model prediction information storage area, and notify prediction command reception module 131 carry out prediction processing, internal structure is shown in Fig. 9.
Model training module 124 is responsible for receiving the model training request from feature processing block 123, stores mould from data
Model training information storage area, the preset rules memory block of block 112 obtain model training sample data and preset model instruction respectively
Practice rule, carries out model training using machine learning algorithm.Wherein, machine learning algorithm selects DRF random forests algorithm, this calculation
Method is a kind of decision Tree algorithms of comparative maturity, and details are not described herein for concrete principle, and model training is tested using N-cross intersection
Card, after model training threshold value, which reaches preset rules, to be required, by the training of training pattern data storage to data memory module 112
Model storage area, and return to training result.
Fig. 5 prediction meanss 13 include prediction command reception module 131, on-line prediction processing module 132, offline prediction processing
Module 133 and prediction result output module 134, in which:
Predict command reception module 131 be responsible for receive from peripheral system, model training apparatus 12, marketing apparatus 14 it is pre-
It surveys and instructs and notify corresponding prediction command process module.Wherein, prediction instruction is initiated offline pre- comprising timing batch system
Survey instruction, the on-line prediction that client's marketing module 141 is initiated instructs, the on-line/off-line of the initiation of customer manager's marketing module 142 is pre-
The on-line/off-line for surveying instruction and the initiation of feature processing block 123 predicts instruction, and on-line prediction is mainly directed towards client and provides the second
Grade real-time recommendation service, processing data volume is small, and response timeliness is high;Offline prediction is mainly directed towards customer manager and provides particular customer group
Or the batch recommendation service of product set, processing data volume is big, and response timeliness is low, and prediction command process module includes on-line prediction
Processing module 132 and offline prediction processing module 133.
On-line prediction processing module 132 is responsible for processing on-line prediction instruction, it is offline predict processing module 133 be responsible for processing from
Line prediction instruction.After receiving instruction, the two modules from data memory module 112 obtain newest client's forecast sample data and
Model data gives a mark in conjunction with preset model prediction rule, and notifies prediction result output module 134.
Prediction result output module 134 is responsible for notification data memory module 112 and completes model prediction information and training pattern
Data update, and prediction result is notified to client's marketing module 141, customer manager's marketing module 142 and characteristic processing mould
Block 123.
Fig. 6 marketing apparatus 14 includes client's marketing module 141 and customer manager's marketing module 142, in which:
Client's marketing module 141 provides the intelligently pushing service of legal person's finance product towards the outer client of row, receives row guest who is not a relative
Family legal person manages money matters after proposal inquiry request, initiates on-line prediction request to prediction command reception module 131, stores mould by data
Block 112 obtains final prediction result, and the peripheral system such as Web bank, enterprise 8, Mobile banking 9 etc. for being pushed to direct-connected client.
Meanwhile client's marketing module 141 docks peripheral system, completes client's essential information, customer transactional data, System Operation Log etc.
Real-time data acquisition, out-out business system includes but is not limited to enterprise-level customer information system 2, to public client's basic service herein
System 3, credit system 5, legal person's marketing system 6, financing system 7, Web bank, enterprise 8, Mobile banking 9 etc..
Customer manager's marketing module 142 provides the intelligently pushing service of legal person's finance product towards customer manager in row, described
Intelligent marketing service includes model training, legal person manages money matters, and scheme exports, preset rules adjustment and marketing data are collected.Wherein:
Model training, which is responsible for processing, requests public customer manager's model real-time update, and notice training instruction processing module 121 initiates model
Training managing.Legal person's scheme output of managing money matters is responsible for processing and manages money matters solution formulation request to public customer manager legal person, by prediction
Command reception module 131 initiates on-line/off-line predictions request, obtains final prediction result from data memory module 112, and push
To customer manager's marketing platform such as credit system 4, corporate client's marketing system 6 etc..Preset rules adjustment can be according to client and visitor
Handle the parameters such as actual demand setting data processing rule, model training rule, model prediction rule, feature Expert Rules in family.
Marketing data, which is collected, is responsible for completing the off-line data collectings such as customer manager's marketing data in row, marketing log.
Fig. 7 data acquisition module 111 includes the acquisition of client's natural information, customer action information collection, finance product information
Acquisition, four part of marketing of bank information collection, this four modules are sought by Kafka from client's marketing module 141 and customer manager
Module 142 is sold to acquire basic data and store to data memory module 112.This four partial information with client bank Customer ID
It being attached for major key, data acquisition module 111 acquires nearly 1 year customer action information of client, marketing of bank information, and
It manages money matters to public client legal person and buys the client's natural information and finance product information of time point, acquisition matrix table such as the following table 1:
1 data acquisition matrix table of table
Fig. 8 data memory module 112 includes acquired data storage area, model training information storage area, model prediction information
Memory block, training pattern memory block, five part of preset rules memory block, in which:
It is responsible for storing the basis acquisition data from data acquisition module 111, model training information in acquired data storage area
The model training sample after characteristic processing is responsible for storing in memory block, and model prediction information storage area is responsible for storing at through feature
Storage model training is responsible in model prediction sample data and on-line/off-line prediction result after reason, training pattern memory block
The model information that module 124 exports, it is existing that preset rules store rules, the uses such as responsible storing data processing, model training
Hadoop big data platform completes data storage.Wherein preset rules sample table such as the following table 2:
2 preset rules sample table of table
Fig. 9 feature processing block 123 includes characteristic processing command reception, aspect of model processing, the aspect of model extracts and place
It manages result and exports four parts.Wherein:
Characteristic processing command reception Interworking Data processing module 122 is responsible for receiving characteristic processing request;Aspect of model processing
It is responsible for obtaining model training data or model prediction information from data memory module 112, in conjunction with default feature Expert Rules, processing
Generate final characteristic set.Aspect of model extraction is responsible on the basis of final characteristic set, has with machine learning algorithm extraction
Imitate characteristic set, and by processing result output module by after processing data and characteristic set storage to data memory module
112, specific process flow is shown in Figure 11.
Figure 10 is a kind of working-flow figure of legal person's financing intelligent recommendation system of the present invention, as shown in Figure 10, legal person
Finance product intelligently pushing system includes inquiry financing suggested design and initiation two sub- workflows of model training, the two sub- works
The same data storage area is shared as stream.Wherein, initiating model training workflow is legal person's finance product intelligently pushing system work
Make basis, which can be started automatically by timing batch system, can also be marketed by customer manager in row by customer manager
142 start-up by hand of module;Inquiry financing suggested design workflow is carried out based on the model training result of model training workflow, should
Workflow can be started automatically by timing batch system, can also pass through 142 hand of client's marketing module 141 or customer manager's marketing module
Work starting.
It is as follows to initiate the sub- workflow process steps of model training:
Step S00: the model training instruction for coming self-timing batch system or customer manager's marketing module 142, root are received
According to command content judge in data memory module 112 whether existing available model, and if so, directly notice timing batch
System or the instruction of customer manager's marketing module 142 are handled successfully, otherwise Boot Model training process.Herein, command content packet
It includes positive/negative sample and defines information, model training algorithm, model training rule etc., with the well-informed legal person's financing of Xiang little Wei lead referral e
Be illustrated for product: small micro- client buys the well-informed legal person's finance product of e in the model training positive sample half a year of being defined as over
Detail, each purchaser record is a positive sample, and model training negative sample, which is defined as over, did not bought the well-informed method of e in 1 year
The client of people's finance product, positive and negative sample proportion are set as 1:30, and training set, test set and verifying integrate ratio setting as 1:1:3,
And it is 0.8 that model training threshold value F1, which is arranged,.The standard for judging whether there is available model is that the model of data memory module 112 is deposited
The training pattern and final updating date > current date-batch forecast frequency of the existing same type of storage area.
Step S01: after training instruction processing module 121 receives model training instruction, decision instruction type: if it is next
The batch training instruction of self-timing batch system, then the preset model training rules of ergodic data memory module 112, notify one by one
Data acquisition module 111 obtains the basic datas information such as newest transaction data, marketing log, and stores and arrive data memory module
112 acquired data storage area, data type are that training sample acquires data, and notification data processing module 122 carries out data
Working process;If it is the on-line training instruction from customer manager's marketing module 142, data storage is searched for according to command content
The preset model training rules of module 112, if not being matched to record, a newly-increased model training rule is deposited to preset rules
Storage area, follow-up processing flow is the same as batch training instruction.
Step S02: after data processing module 122 receives data mart modeling instruction, from the acquisition number of data memory module 112
Training sample data are obtained according to memory block, obtain data processing rule from the preset rules memory block of data memory module 112, it is right
Training sample data are processed, and feature processing block 123 is notified to carry out feature machining.The data processing rule packet
Include but be not limited to dealing of abnormal data, amount of money Logarithm conversion, the conversion of date integer etc..
Step S03: feature processing block 123 receive feature machining instruction after, to training sample feature carry out processing and
Processing, obtains final characteristic set, and updates model training sample data to data memory module 112, notification model training mould
Block 124 carries out model training processing, and the detailed process of characteristic processing is shown in Figure 11.
Step S04: after model training module 124 receives model training instruction, from the training number of data memory module 112
Training sample data are obtained according to memory block, model training rule is obtained from the preset rules memory block of data memory module 112, opens
Movable model trains process, if model training threshold value is not up to preset threshold, returns to model training failure, batch training is referred to
It enables, active push failure information to customer manager's marketing module 142, failure information includes model details and presets herein
Model training rule etc.;Reach preset threshold and then update model data to the model storage area of data memory module 112, and returns
Model training success.
The inquiry financing sub- workflow process steps of suggested design are as follows:
Step S10: the prediction instruction for coming self-timing batch system, model training apparatus 12 and marketing apparatus 14 is received
Afterwards, starting prediction process flow.The prediction instruction includes the offline prediction instruction for carrying out self-timing batch system, from client battalion
Sell the on-line prediction instruction of module 141, the on-line/off-line prediction instruction from customer manager's marketing module 142 and from spy
The on-line/off-line for levying processing module 123 predicts instruction.
Step S11: prediction command reception module 131 receive prediction instruction after, to the prediction command content received into
Row judgement, the prediction command content includes prediction instruction type, prediction customer information, prediction product information, wherein prediction visitor
Family information includes but is not limited to Customer ID, customer group's type etc., and prediction product information includes but is not limited to product IDs, product class
Type etc..If instruction type is on-line prediction, notice on-line prediction processing module 132 is handled, if instruction type is offline pre-
It surveys, notifies that offline prediction processing module 133 is handled.On-line prediction processing module 132 is mainly used for corporation sole client's reason
Wealth scheme is recommended, and is illustrated process flow for predicting that certain buys target finance product to public client: on-line prediction processing
Module 132 obtains the newest pre- test sample of client from the model prediction information storage area of data memory module 112 according to prediction Customer ID
Notebook data is directly entered step S14 if Customer ID exists;Explanation is new client if Customer ID is not present, and need to pass through data
Acquisition device 11 carries out data acquisition and storage, and notification data processing module 122 carries out data mart modeling processing.At offline prediction
Reason module 133 is mainly used for the financing scheme output of customer group and product set, such as predicts that medium-sized and small enterprises buy specific reason
The probability of property product, the target customer for excavating new online finance product etc., detailed process and 132 phase of on-line prediction processing module
Together, details are not described herein.
Step S12: after data processing module 122 receives the instruction of forecast sample data mart modeling, from data memory module 112
Acquired data storage area obtain forecast sample acquire data, from the preset rules memory block of data memory module 112 obtain number
According to processing rule, forecast sample acquisition data are processed, and feature processing block 123 is notified to carry out feature machining.
Step S13: feature processing block 122 receive feature machining instruction after, to forecast sample feature carry out processing and
Processing, obtains final characteristic set, and update forecast sample data to data memory module 112, notice prediction command reception mould
Block 131 carries out prediction processing, and the detailed process of characteristic processing is shown in Figure 11.
Step S14: on-line prediction processing module 132 and offline prediction processing module 133 receive after predicting process instruction,
Newest forecast sample data are obtained from the prediction data memory block of data memory module 112, from the model of data memory module 112
Memory block obtains updated model, model prediction rule is obtained from the preset rules memory block of data memory module 112, to pre- test sample
This carries out prediction scoring, updates the model prediction information storage area of data to data memory module 112, data here include mould
Type forecast sample data, each forecast sample score information and prediction result, newest model data etc., data, which update, to be completed
Afterwards, and recommendation results are exported to marketing apparatus 14.
Step S15: the legal person that client's marketing module 141 receives model prediction device 13 manages money matters after recommendation results, to right
Objective operation system output financing suggested design, described here includes but is not limited to financing system 7, in enterprise network to objective operation system
Bank 8, Mobile banking 9 etc., the financing suggested design particular content include Customer ID, recommend finance product ID, purchase probability,
Client's purchase decision supports information etc., and wherein client's purchase decision support information includes client's current account balance, clients fund
Plan, the currently available share of finance product etc..
Step S16: the legal person that customer manager's marketing module 142 receives model prediction device 13 manages money matters after recommendation results,
Into row, operation system exports financing suggested design, and operation system includes but is not limited to credit system 5, legal person in row described here
Marketing system 6, financing system 7 etc., the financing suggested design particular content include Customer ID group, recommend finance product ID collection
Conjunction, purchase probability, marketing decision-making support information etc., wherein marketing decision-making support information includes client characteristics set, client's financing
Buy detail, finance product transaction statistical data etc..
Step S17: after each out-out business system receives financing suggested design, generating business transaction data, marketing log,
This partial data is distributed in existing business system as the basic data of model training and prediction, by client's marketing module 141
Acquisition and centrally stored, specific visible Fig. 7 and Fig. 8 are completed with customer manager's marketing module 142.
Figure 11 is feature of present invention process flow diagram, and as shown in figure 11, characteristic processing step includes:
Step S100: after receiving the characteristic processing request from data processing module 122, data memory module is read
Acquisition data in 112, if latest data date≤current date-batch forecast frequency or latest data and last time
Data variation rate is more than 10%, then starting step S101.Wherein, data variation rate=| 1- existing customer number/last time client's number |.
Step S101: after reading newest acquisition data from the acquired data storage area in data memory module 112, in conjunction with
Default Expert Rules carry out preliminary screening to basic data, remove single data column, and single data column refer to all here
Client shows unconverted feature.
Step S102: in conjunction with the foundation characteristic after step S101 screening, middle level features processing, middle level features here are carried out
It mainly include customer action temporal aspect and product variations temporal aspect two parts.Wherein, customer action temporal aspect is to visitor
The behavior of family for a period of time is analyzed, and processing obtains temporal aspect, and by taking buying behavior of managing money matters as an example, client's temporal aspect includes
Finance product is bought in the nearly three months/half a year/1 year purchase finance product stroke count & amount of money of client, nearly three months/half a year/1 year of client
Channel etc.;Product variations temporal aspect is then bright to the earning rate of a period of time, risk class, sale after finance product distribution
Thin wait carries out Machining Analysis, obtains temporal aspect, highest/minimum/average receipts such as target financing in nearly three months/half a year/1 year
Beneficial rate, target are managed money matters highest/minimum/average sale etc. of nearly three months/half a year/1 year.
Step S103: the foundation characteristic of the middle level features and S101 step screening that obtain in conjunction with S102 step processing carries out advanced
Feature construction.Here advanced features construct legal person's financing customer portrait feature mainly from customer perspective.Mainly include:
Client's liveness, client's channel preference, client financing risk partiality, clients fund retention ratio, settlement bargain concentration degree, Ke Hurong
Provide risk factor etc..
Step S104: after S101, S102, S103 walk feature extraction and complete the process, PCA Feature Dimension Reduction is used first
Technology rejects uncorrelated features, if characteristic dimension is still excessively high, automated characterization screening implement boruta can be used to carry out feature
Selection.Boruta and PCA is existing characteristic processing algorithm, and details are not described herein.
Step S105: after the completion of Feature Selection, if what is received is training characteristics processing request, by the data after processing
It stores with characteristic set to the model training information storage area of data memory module 112, and notification model training module 124 carries out
Final mask training;If what is received is predicted characteristics processing request, by the data storage after processing to data memory module
112 model prediction information storage area, and prediction command reception module 131 is notified to carry out prediction processing.
Legal person's finance product intelligently pushing system and method provided by the invention, efficiently solves business bank to public client
The problems such as finance product marketing is difficult, at high cost, covering surface is narrow, has the effect of following and advantage:
(1), fast accurate is marketed.Real-time reception client financing recommendation request, collects online and handles data, pre- in real time
Assessment point, quickly exports recommendation results;Support that target customers or target product are carried out by preset rules to be screened, and is reached simultaneously
Then pass through business bank's marketing platform to preset threshold and be pushed to client, realizes precision marketing.
(2), marketing view abundant is provided.On the basis of data collection, storage, client is constructed for corporate client and is drawn
Picture exports complete customer view;Export newest financing information in real time and client hold financing history detail, be client and
Customer manager provides information to aid in decision abundant.
(3), long-tail client is covered comprehensively.Bank can be covered comprehensively in a manner of machine substitute human labor to public client.Effectively excavate
Potentiality client and wake-up are sunk into sleep client, and marketing human input is saved, and are reduced customer manager and are tieed up objective cost.
System, device, module or the unit that above-described embodiment illustrates can specifically realize by computer chip or entity,
Or it is realized by the product with certain function.It is a kind of typically to realize that equipment is computer equipment, specifically, computer is set
It is standby for example can for personal computer, laptop computer, cellular phone, camera phone, smart phone, personal digital assistant,
Media player, navigation equipment, electronic mail equipment, game console, tablet computer, wearable device or these equipment
In any equipment combination.
Computer equipment specifically includes memory, processor and storage on a memory simultaneously in a typical example
The computer program that can be run on a processor is realized when the processor executes described program and is held as described above by client
Capable method, alternatively, the processor realizes the method executed as described above by server when executing described program.
Below with reference to Figure 18, it illustrates the structures for the computer equipment 600 for being suitable for being used to realize the embodiment of the present application to show
It is intended to.
As shown in figure 18, computer equipment 600 includes central processing unit (CPU) 601, can be read-only according to being stored in
Program in memory (ROM) 602 is loaded into random access storage device (RAM) from storage section 608) program in 603
And execute various work appropriate and processing.In RAM603, also it is stored with system 600 and operates required various program sum numbers
According to.CPU601, ROM602 and RAM603 are connected with each other by bus 604.Input/output (I/O) interface 605 is also connected to
Bus 604.
I/O interface 605 is connected to lower component: the importation 606 including keyboard, mouse etc.;It is penetrated including such as cathode
The output par, c 607 of spool (CRT), liquid crystal display (LCD) etc. and loudspeaker etc.;Storage section 608 including hard disk etc.;
And including such as LAN card, the communications portion 609 of the network interface card of modem etc..Communications portion 609 via such as because
The network of spy's net executes communication process.Driver 610 is also connected to I/O interface 606 as needed.Detachable media 611, such as
Disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on as needed on driver 610, in order to read from thereon
Computer program be mounted as needed such as storage section 608.
Particularly, according to an embodiment of the invention, may be implemented as computer above with reference to the process of flow chart description
Software program.For example, the embodiment of the present invention includes a kind of computer program product comprising be tangibly embodied in machine readable
Computer program on medium, the computer program include the program code for method shown in execution flow chart.At this
In the embodiment of sample, which can be downloaded and installed from network by communications portion 609, and/or from removable
Medium 611 is unloaded to be mounted.
Computer-readable medium includes permanent and non-permanent, removable and non-removable media can be by any method
Or technology come realize information store.Information can be computer readable instructions, data structure, the module of program or other data.
The example of the storage medium of computer includes, but are not limited to phase change memory (PRAM), static random access memory (SRAM), moves
State random access memory (DRAM), other kinds of random access memory (RAM), read-only memory (ROM), electric erasable
Programmable read only memory (EEPROM), flash memory or other memory techniques, read-only disc read only memory (CD-ROM) (CD-ROM),
Digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape magnetic disk storage or other magnetic storage devices
Or any other non-transmission medium, can be used for storage can be accessed by a computing device information.As defined in this article, it calculates
Machine readable medium does not include temporary computer readable media (transitory media), such as the data-signal and carrier wave of modulation.
For convenience of description, it is divided into various units when description apparatus above with function to describe respectively.Certainly, implementing this
The function of each unit can be realized in the same or multiple software and or hardware when application.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
It should also be noted that, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, commodity or the equipment that include a series of elements not only include those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, commodity or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including described want
There is also other identical elements in the process, method of element, commodity or equipment.
It will be understood by those skilled in the art that embodiments herein can provide as method, system or computer program product.
Therefore, complete hardware embodiment, complete software embodiment or embodiment combining software and hardware aspects can be used in the application
Form.It is deposited moreover, the application can be used to can be used in the computer that one or more wherein includes computer usable program code
The shape for the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
Formula.
The application can describe in the general context of computer-executable instructions executed by a computer, such as program
Module.Generally, program module includes routines performing specific tasks or implementing specific abstract data types, programs, objects, group
Part, data structure etc..The application can also be practiced in a distributed computing environment, in these distributed computing environments, by
Task is executed by the connected remote processing devices of communication network.In a distributed computing environment, program module can be with
In the local and remote computer storage media including storage equipment.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for system reality
For applying example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to embodiment of the method
Part explanation.
The above description is only an example of the present application, is not intended to limit this application.For those skilled in the art
For, various changes and changes are possible in this application.All any modifications made within the spirit and principles of the present application are equal
Replacement, improvement etc., should be included within the scope of the claims of this application.
Claims (24)
1. a kind of finance product intelligently pushing method characterized by comprising
Obtain the identity of client to be pushed;
By client's archive information of the correspondence identity and the preset prediction model of customer action information input, managed money matters
The push of product is as a result, the prediction model is obtained by forecast sample training;
Push the push result;
Wherein, the forecast sample includes that the financing of client's archive information, customer action information, the client selection of multiple clients produces
Product information and current bank marketing message.
2. finance product intelligently pushing method according to claim 1, which is characterized in that client's archive information includes:
Client opens an account information, customer capital information and credit information.
3. finance product intelligently pushing method according to claim 1, which is characterized in that the customer action information includes:
Behavioural information of financing and trading activity information.
4. finance product intelligently pushing method according to claim 1, which is characterized in that the finance product information includes:
Product risks rank, issuing way and expected yield.
5. finance product intelligently pushing method according to claim 1, which is characterized in that further include:
Establish initial predicted model;
Pass through the forecast sample training initial predicted model.
6. finance product intelligently pushing method according to claim 1, which is characterized in that described to correspond to the identity
Client's archive information and the preset prediction model of customer action information input, comprising:
Characteristic processing is carried out the client's archive information and customer action information of the correspondence identity, generates corresponding spy
Sign;
The feature is input to the prediction model, wherein the forecast sample is believed by client's archive information, customer action
The feature set formed after finance product information and current bank marketing message characteristic processing that breath, client select.
7. finance product intelligently pushing method according to claim 6, which is characterized in that the characteristic processing of the forecast sample
The step of include:
The up-to-date information set in duration is differed with current time from screening in the information to characteristic processing;
The up-to-date information is screened, removal obtains foundation characteristic after showing unconverted information to all clients;
The corresponding foundation characteristic of the customer action information is analyzed, customer action temporal aspect is obtained;
The corresponding foundation characteristic of the finance product information is analyzed, product variations temporal aspect is obtained;
According to the customer action temporal aspect and the product variations temporal aspect, customer portrait feature, the client are constructed
Figure Characteristics include: client's liveness, client's channel preference, client financing risk partiality, clients fund retention ratio, settlement bargain
Concentration degree and client's financing risk coefficient.
8. finance product intelligently pushing method according to claim 7, which is characterized in that the characteristic processing of the forecast sample
The step of further include:
Reject the feature that the degree of correlation in the customer portrait feature is lower than given threshold;
Feature Dimension Reduction processing is carried out to the customer portrait feature after rejecting.
9. finance product intelligently pushing method according to claim 5, which is characterized in that described to pass through forecast sample training institute
State initial predicted model, comprising:
Called data processing rule and model training rule;
Data processing is carried out to the forecast sample by the data processing rule;
To data, treated that the forecast sample carries out characteristic processing, obtains characteristic set;
By the feature in conjunction with the initial predicted model is input to, according to the model training rule training initial predicted mould
Type;
Grey iterative generation model training threshold value is updated described pre- based on the size relation of the model training threshold value and preset threshold
Test sample sheet is replaced former forecast sample with the forecast sample of update, until model training threshold value is greater than preset threshold, and then is obtained
The prediction model completed to training.
10. finance product intelligently pushing method according to claim 9, which is characterized in that the data processing rule includes:
Dealing of abnormal data, amount of money Logarithm conversion, the conversion of date integer;
The model training rule includes: training algorithm, positive and negative sample proportion, training set and verifying collection ratio, frequency of training.
11. the matching degree of a kind of client and finance product determines method characterized by comprising
It obtains client's archive information of client to be matched, customer action information, the product information of finance product to be matched and works as
Preceding marketing of bank information;
Client's archive information, customer action information, the product information of finance product to be matched and current bank are marketed
The preset prediction model of information input generates the matching degree of client to be matched Yu the finance product to be matched;
Wherein, the prediction model is obtained by forecast sample training, and the forecast sample includes that the client of multiple clients deposits
Shelves information, customer action information, the finance product information and current bank marketing message of client's selection.
12. matching degree determines method according to claim 11, which is characterized in that client's archive information includes: client
It opens an account information, customer capital information and credit information.
13. matching degree determines method according to claim 11, which is characterized in that the customer action information includes: financing
Behavioural information and trading activity information.
14. matching degree determines method according to claim 11, which is characterized in that the finance product information includes: product
Risk class, issuing way and expected yield.
15. matching degree determines method according to claim 11, which is characterized in that further include:
Establish initial predicted model;
Pass through the forecast sample training initial predicted model.
16. matching degree determines method according to claim 11, which is characterized in that will be by client's archive information, client
Behavioural information, the product information of finance product to be matched and current bank marketing message, comprising:
It markets to client's archive information, customer action information, the product information of finance product to be matched and current bank
Information carries out characteristic processing, generates corresponding feature;
The feature is input to the prediction model, wherein the forecast sample is believed by client's archive information, customer action
The feature set formed after finance product information and current bank marketing message characteristic processing that breath, client select.
17. 6 matching degrees determine method according to claim 1, which is characterized in that the step of the characteristic processing of the forecast sample
Suddenly include:
The up-to-date information set in duration is differed with current time from screening in the information to characteristic processing;
The up-to-date information is screened, removal obtains foundation characteristic after showing unconverted information to all clients;
The corresponding foundation characteristic of the customer action information is analyzed, customer action temporal aspect is obtained;
The corresponding foundation characteristic of the finance product information is analyzed, product variations temporal aspect is obtained;
According to the customer action temporal aspect and the product variations temporal aspect, customer portrait feature, the client are constructed
Figure Characteristics include: client's liveness, client's channel preference, client financing risk partiality, clients fund retention ratio, settlement bargain
Concentration degree and client's financing risk coefficient.
18. 7 matching degrees determine method according to claim 1, which is characterized in that the step of the characteristic processing of the forecast sample
Suddenly further include:
Reject the feature that the degree of correlation in the customer portrait feature is lower than given threshold;
Feature Dimension Reduction processing is carried out to the customer portrait feature after rejecting.
19. matching degree determines method according to claim 15, which is characterized in that described described just by forecast sample training
Beginning prediction model, comprising:
Called data processing rule and model training rule;
Data processing is carried out to the forecast sample by the data processing rule;
To data, treated that the forecast sample carries out characteristic processing, obtains characteristic set;
By the feature in conjunction with the initial predicted model is input to, according to the model training rule training initial predicted mould
Type;
Grey iterative generation model training threshold value is updated described pre- based on the size relation of the model training threshold value and preset threshold
Test sample sheet is replaced former forecast sample with the forecast sample of update, until model training threshold value is greater than preset threshold, and then is obtained
The prediction model completed to training.
20. 9 matching degrees determine method according to claim 1, which is characterized in that the data processing rule includes:
Dealing of abnormal data, amount of money Logarithm conversion, the conversion of date integer;
The model training rule includes: training algorithm, positive and negative sample proportion, training set and verifying collection ratio, frequency of training.
21. a kind of finance product intelligently pushing system characterized by comprising
First obtains module, obtains the identity of client to be pushed;
Result-generation module is pushed, client's archive information of the correspondence identity and customer action information input is preset
Prediction model obtains the push of finance product as a result, the prediction model is obtained by forecast sample training;
Pushing module pushes the push result;
Wherein, the forecast sample includes that the financing of client's archive information, customer action information, the client selection of multiple clients produces
Product information and current bank marketing message.
22. the matching degree of a kind of client and finance product determines system characterized by comprising
Second obtains module, obtains the production of the client's archive information, customer action information, finance product to be matched of client to be matched
Product information and current bank marketing message;
Matching module, by client's archive information, customer action information, the product information of finance product to be matched and current
The preset prediction model of marketing of bank information input generates the matching degree of client to be matched Yu the finance product to be matched;
Wherein, the prediction model is obtained by forecast sample training, and the forecast sample includes that the client of multiple clients deposits
Shelves information, customer action information, the finance product information and current bank marketing message of client's selection.
23. a kind of computer equipment including memory, processor and stores the meter that can be run on a memory and on a processor
Calculation machine program, which is characterized in that the processor realizes the described in any item methods of claim 1 to 20 when executing described program
The step of.
24. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program quilt
The step of claim 1 to 20 described in any item methods are realized when processor executes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910583510.9A CN110276668A (en) | 2019-07-01 | 2019-07-01 | The method and system that finance product intelligently pushing, matching degree determine |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910583510.9A CN110276668A (en) | 2019-07-01 | 2019-07-01 | The method and system that finance product intelligently pushing, matching degree determine |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110276668A true CN110276668A (en) | 2019-09-24 |
Family
ID=67963797
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910583510.9A Pending CN110276668A (en) | 2019-07-01 | 2019-07-01 | The method and system that finance product intelligently pushing, matching degree determine |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110276668A (en) |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110766462A (en) * | 2019-10-23 | 2020-02-07 | 中国工商银行股份有限公司 | Intelligent panoramic customer portrait linkage method and system based on streaming platform |
CN111340546A (en) * | 2020-02-25 | 2020-06-26 | 中信银行股份有限公司 | Method, device, computer equipment and readable storage medium for improving marketing efficiency of banking business |
CN111523676A (en) * | 2020-04-17 | 2020-08-11 | 第四范式(北京)技术有限公司 | Method and device for assisting machine learning model to be online |
CN111626766A (en) * | 2020-04-23 | 2020-09-04 | 深圳索信达数据技术有限公司 | Mobile banking marketing customer screening method integrating multiple machine learning models |
CN111695938A (en) * | 2020-06-05 | 2020-09-22 | 中国工商银行股份有限公司 | Product pushing method and system |
CN111859922A (en) * | 2020-07-31 | 2020-10-30 | 上海银行股份有限公司 | Application method of entity relation extraction technology in bank wind control |
CN111861727A (en) * | 2020-07-31 | 2020-10-30 | 平安普惠企业管理有限公司 | Data distribution method, device, computer equipment and storage medium |
CN111882113A (en) * | 2020-07-01 | 2020-11-03 | 中国银行股份有限公司 | Enterprise mobile banking user prediction method and device |
CN111951113A (en) * | 2020-08-17 | 2020-11-17 | 中国银行股份有限公司 | Investment service method and device based on big data |
CN112150290A (en) * | 2020-09-08 | 2020-12-29 | 广元量知汇科技有限公司 | Product pushing method for intelligent finance |
CN112288447A (en) * | 2020-10-30 | 2021-01-29 | 北京每日优鲜电子商务有限公司 | Article information display method and device, electronic equipment and computer readable medium |
CN112580917A (en) * | 2019-09-30 | 2021-03-30 | 深圳无域科技技术有限公司 | Client characteristic evaluation method and device |
CN112750043A (en) * | 2021-01-14 | 2021-05-04 | 中国工商银行股份有限公司 | Business data pushing method and device and server |
CN113129024A (en) * | 2021-05-19 | 2021-07-16 | 中国银行股份有限公司 | Intelligent financing method and system based on 5G message of block chain |
CN113344709A (en) * | 2021-06-23 | 2021-09-03 | 中国银行股份有限公司 | Client financing risk assessment method and device |
CN113450158A (en) * | 2021-07-09 | 2021-09-28 | 中国银行股份有限公司 | Bank activity information pushing method and device |
CN114926287A (en) * | 2022-06-02 | 2022-08-19 | 河北源达信息技术股份有限公司 | Method and device for intelligently recommending investment products and electronic equipment |
TWI829404B (en) * | 2022-10-28 | 2024-01-11 | 第一商業銀行股份有限公司 | Intelligent time series marketing system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106600369A (en) * | 2016-12-09 | 2017-04-26 | 广东奡风科技股份有限公司 | Real-time recommendation system and method of financial products of banks based on Naive Bayesian classification |
CN109447728A (en) * | 2018-09-07 | 2019-03-08 | 平安科技(深圳)有限公司 | Financial product recommended method, device, computer equipment and storage medium |
CN109767345A (en) * | 2018-12-15 | 2019-05-17 | 深圳壹账通智能科技有限公司 | Products Show method, apparatus, equipment and storage medium based on information matches |
CN109934721A (en) * | 2019-01-18 | 2019-06-25 | 深圳壹账通智能科技有限公司 | Finance product recommended method, device, equipment and storage medium |
-
2019
- 2019-07-01 CN CN201910583510.9A patent/CN110276668A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106600369A (en) * | 2016-12-09 | 2017-04-26 | 广东奡风科技股份有限公司 | Real-time recommendation system and method of financial products of banks based on Naive Bayesian classification |
CN109447728A (en) * | 2018-09-07 | 2019-03-08 | 平安科技(深圳)有限公司 | Financial product recommended method, device, computer equipment and storage medium |
CN109767345A (en) * | 2018-12-15 | 2019-05-17 | 深圳壹账通智能科技有限公司 | Products Show method, apparatus, equipment and storage medium based on information matches |
CN109934721A (en) * | 2019-01-18 | 2019-06-25 | 深圳壹账通智能科技有限公司 | Finance product recommended method, device, equipment and storage medium |
Non-Patent Citations (1)
Title |
---|
郭福春,等主编: "《互联网金融概论 第2版》", 30 September 2018, 中国金融出版社 * |
Cited By (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112580917A (en) * | 2019-09-30 | 2021-03-30 | 深圳无域科技技术有限公司 | Client characteristic evaluation method and device |
CN112580917B (en) * | 2019-09-30 | 2024-04-05 | 深圳无域科技技术有限公司 | Customer characteristic evaluation method and device |
CN110766462A (en) * | 2019-10-23 | 2020-02-07 | 中国工商银行股份有限公司 | Intelligent panoramic customer portrait linkage method and system based on streaming platform |
CN111340546A (en) * | 2020-02-25 | 2020-06-26 | 中信银行股份有限公司 | Method, device, computer equipment and readable storage medium for improving marketing efficiency of banking business |
CN111523676A (en) * | 2020-04-17 | 2020-08-11 | 第四范式(北京)技术有限公司 | Method and device for assisting machine learning model to be online |
CN111523676B (en) * | 2020-04-17 | 2024-04-12 | 第四范式(北京)技术有限公司 | Method and device for assisting machine learning model to be online |
CN111626766A (en) * | 2020-04-23 | 2020-09-04 | 深圳索信达数据技术有限公司 | Mobile banking marketing customer screening method integrating multiple machine learning models |
CN111695938A (en) * | 2020-06-05 | 2020-09-22 | 中国工商银行股份有限公司 | Product pushing method and system |
CN111695938B (en) * | 2020-06-05 | 2023-07-18 | 中国工商银行股份有限公司 | Product pushing method and system |
CN111882113A (en) * | 2020-07-01 | 2020-11-03 | 中国银行股份有限公司 | Enterprise mobile banking user prediction method and device |
CN111882113B (en) * | 2020-07-01 | 2024-02-27 | 中国银行股份有限公司 | Enterprise mobile banking user prediction method and device |
CN111859922B (en) * | 2020-07-31 | 2023-12-01 | 上海银行股份有限公司 | Application method of entity relation extraction technology in bank wind control |
CN111861727A (en) * | 2020-07-31 | 2020-10-30 | 平安普惠企业管理有限公司 | Data distribution method, device, computer equipment and storage medium |
CN111859922A (en) * | 2020-07-31 | 2020-10-30 | 上海银行股份有限公司 | Application method of entity relation extraction technology in bank wind control |
CN111951113A (en) * | 2020-08-17 | 2020-11-17 | 中国银行股份有限公司 | Investment service method and device based on big data |
CN112150290A (en) * | 2020-09-08 | 2020-12-29 | 广元量知汇科技有限公司 | Product pushing method for intelligent finance |
CN112288447A (en) * | 2020-10-30 | 2021-01-29 | 北京每日优鲜电子商务有限公司 | Article information display method and device, electronic equipment and computer readable medium |
CN112750043A (en) * | 2021-01-14 | 2021-05-04 | 中国工商银行股份有限公司 | Business data pushing method and device and server |
CN112750043B (en) * | 2021-01-14 | 2024-02-02 | 中国工商银行股份有限公司 | Service data pushing method, device and server |
CN113129024A (en) * | 2021-05-19 | 2021-07-16 | 中国银行股份有限公司 | Intelligent financing method and system based on 5G message of block chain |
CN113344709A (en) * | 2021-06-23 | 2021-09-03 | 中国银行股份有限公司 | Client financing risk assessment method and device |
CN113450158A (en) * | 2021-07-09 | 2021-09-28 | 中国银行股份有限公司 | Bank activity information pushing method and device |
CN114926287A (en) * | 2022-06-02 | 2022-08-19 | 河北源达信息技术股份有限公司 | Method and device for intelligently recommending investment products and electronic equipment |
TWI829404B (en) * | 2022-10-28 | 2024-01-11 | 第一商業銀行股份有限公司 | Intelligent time series marketing system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110276668A (en) | The method and system that finance product intelligently pushing, matching degree determine | |
Chursin et al. | Innovation as a Basis for Competitiveness | |
US20180129961A1 (en) | System, method and computer-accessible medium for making a prediction from market data | |
CA2489414A1 (en) | System and method for estimating and optimizing transaction costs | |
CN112053097A (en) | Loan collection method and device, electronic equipment and storage medium | |
US20120053973A1 (en) | Method of repairing financially infeasible genetic algorithm chromosome encoding activity start times in scheduling | |
CN112074860A (en) | Computer-implemented method for compiling an investment portfolio of assets | |
JP2019012518A (en) | Advice generation device, advice presentation system, advice generation program, and advice data generation system | |
JP2018503927A (en) | Segmentation and stratification of composite portfolio of investment securities | |
CN108492189A (en) | Portfolio Selection selection method, device and equipment | |
CN112613997A (en) | Method and apparatus for forecasting combined investment of money fund | |
CN110335090A (en) | Replenishing method and system, electronic equipment based on Sales Volume of Commodity forecast of distribution | |
CN112132639A (en) | Dynamic pricing method of data set based on machine learning | |
Wang et al. | Resource portfolio planning of make-to-stock products using a constraint programming-based genetic algorithm | |
Ronyastra et al. | A multi criteria decision analysis for reinvestment action portfolio selection problem in an Indonesian real estate company | |
US20070033120A1 (en) | System and method for organization of financial structured products | |
CN109767333A (en) | Select based method, device, electronic equipment and computer readable storage medium | |
Janani et al. | Selection of portfolio by using multi attributed decision making (Tehran stock exchange) | |
CN110858355A (en) | Project budget balance prediction method and device | |
JP7560201B2 (en) | Negotiation system, negotiation method and negotiation program | |
KR102176108B1 (en) | Differential fee payment system through professional experts | |
Agrawal et al. | Matching intermediaries for information goods in the presence of direct search: an examination of switching costs and obsolescence of information | |
CN113313562B (en) | Product data processing method and device, computer equipment and storage medium | |
KR101192680B1 (en) | System and method for portfolio management | |
US20230267542A1 (en) | System and method for analysing financial products |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190924 |
|
RJ01 | Rejection of invention patent application after publication |