CN112184304A - Method, system, server and storage medium for assisting decision - Google Patents
Method, system, server and storage medium for assisting decision Download PDFInfo
- Publication number
- CN112184304A CN112184304A CN202011025802.XA CN202011025802A CN112184304A CN 112184304 A CN112184304 A CN 112184304A CN 202011025802 A CN202011025802 A CN 202011025802A CN 112184304 A CN112184304 A CN 112184304A
- Authority
- CN
- China
- Prior art keywords
- target
- decision
- model
- decision model
- assistant
- 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
- 238000000034 method Methods 0.000 title claims abstract description 76
- 230000000694 effects Effects 0.000 claims abstract description 80
- 238000012549 training Methods 0.000 claims abstract description 43
- 238000012216 screening Methods 0.000 claims abstract description 20
- 230000004927 fusion Effects 0.000 claims abstract description 7
- 230000008569 process Effects 0.000 claims description 22
- 230000001502 supplementing effect Effects 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 230000003287 optical effect Effects 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 230000004044 response Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 239000000047 product Substances 0.000 description 3
- 239000013307 optical fiber Substances 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 238000010992 reflux Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000009193 crawling Effects 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000009469 supplementation Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
Images
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
- G06Q30/0255—Targeted advertisements based on user history
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- 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
- G06Q40/02—Banking, e.g. interest calculation or account maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
Abstract
The embodiment of the application discloses a method, a system, a server and a storage medium for assisting decision, wherein the method comprises the following steps: selecting a preset number of target auxiliary decision models from at least three target auxiliary decision models obtained through training, and fusing the selected target auxiliary decision models to obtain a final auxiliary decision model; responding to the trigger operation for starting the target business activity, and screening out target customers from a full-scale customer set by using a pre-trained auxiliary decision model; and pushing the target business activity to the screened target client. In the embodiment of the application, the auxiliary decision model is obtained by adopting a model fusion mode, so that the accuracy of the auxiliary decision model for screening the users is higher, and the auxiliary decision model is used for screening the target clients, so that the problem that business personnel screen the target clients based on business experience is avoided, the business personnel without business experience can efficiently and accurately screen the target clients, and further the relevant activities are accurately pushed to the clients with requirements is solved.
Description
Technical Field
The embodiment of the application relates to the technical field of computers, in particular to a method, a system, a server and a storage medium for assisting decision.
Background
In the financial industry, activities are often promoted to customers to promote customer stickiness. However, due to the need of protecting the customer information, when business personnel conduct activity promotion, the information of the inline seed customer cannot be provided to the advertisement putting companies outside the bank system for customer analysis and accurate advertisement putting. Therefore, business personnel in the financial industry can only capture target customers to be marketed from the database to carry out activity promotion by depending on business experience of the business personnel.
However, this approach has certain disadvantages: the efficiency of target customer screening is low due to the fact that business experience of financial business personnel is excessively depended, once new business occurs or a customer group changes, activities cannot be accurately pushed to needed customers, the effect of activity promotion is poor, and the business personnel need to grope and accumulate experience again.
Disclosure of Invention
The embodiment of the application provides a method, a system, a server and a storage medium for assisting decision-making, so as to achieve the purpose of improving the screening efficiency and accuracy of a target customer when a service worker has no service experience.
In a first aspect, an embodiment of the present application provides a method for assisting decision, where the method includes:
constructing an initial assistant decision model, training the initial assistant decision model in sequence by utilizing a client sample, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models;
selecting a preset number of target assistant decision models from the at least three target assistant decision models, and fusing the selected target assistant decision models to obtain a final assistant decision model;
responding to a trigger operation for starting a target business activity, and screening out a target client from a full client set by using an auxiliary decision model;
and pushing the target business activity to the screened target client.
In a second aspect, an embodiment of the present application further provides a system for assisting decision, where the system includes:
the model training module is used for constructing an initial assistant decision model, sequentially training the initial assistant decision model by utilizing client samples, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models;
the fusion module is used for selecting a preset number of target auxiliary decision models from the at least three target auxiliary decision models and fusing the selected target auxiliary decision models to obtain a final auxiliary decision model;
the auxiliary decision-making module is used for responding to the triggering operation for starting the target business activity and screening out target customers from the full-scale customer set by utilizing an auxiliary decision-making model;
and the pushing module is used for pushing the target business activity to the screened target client.
In a third aspect, an embodiment of the present application further provides a server, including:
one or more processors;
a storage system for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method for assisting decision making as described in any of the embodiments of the present application.
In a fourth aspect, the present application further provides a storage medium, on which a computer program is stored, where the computer program, when executed by a processor, implements a method for assisting decision making according to any one of the embodiments of the present application.
In the embodiment of the application, a plurality of models with better training effects are fused into the assistant decision model in a model fusion mode, so that the accuracy of the assistant decision model for screening users is higher, the trained assistant decision model can be directly utilized to screen out target clients from a full client set in response to the triggering operation of developing target business activities, and then the target business activities are pushed to the target clients. Therefore, the target clients are screened by using the aid of the aid decision model, the condition that a service worker screens the target clients based on service experience is avoided, the service worker without the service experience can efficiently and accurately screen the target clients, and related activities are accurately pushed to clients with requirements.
Drawings
FIG. 1 is a flow chart illustrating a method for assisting decision in one embodiment of the present application;
FIG. 2 is a flowchart illustrating a method for assisting decision-making in a second embodiment of the present application;
FIG. 3 is a flowchart illustrating a method for assisting decision-making in a third embodiment of the present application;
FIG. 4 is a flowchart illustrating a method for assisting decision in a fourth embodiment of the present application;
FIG. 5 is a schematic structural diagram of a system for assisting decision in a fifth embodiment of the present application;
fig. 6 is a schematic structural diagram of a server in a sixth embodiment of the present application.
Detailed Description
The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the application and are not limiting of the application. It should be further noted that, for the convenience of description, only some of the structures related to the present application are shown in the drawings, not all of the structures.
Example one
Fig. 1 is a flowchart of a method for assisting decision according to an embodiment of the present application, where the present embodiment is applicable to a situation where a target customer is accurately screened in a financial scenario and a target business activity is pushed to the target customer, and the method may be executed by a system for assisting decision, where the system may be implemented in a software and/or hardware manner, and may be integrated on a server.
As shown in fig. 1, the method for assisting decision includes the following steps:
s101, responding to a trigger operation for starting target business activities, and screening target customers from a full-scale customer set by using a pre-trained auxiliary decision model.
When some target business activities (such as credit card promotion activities, installment payment activities, and the like) are released by an organization (such as a bank) in the financial industry, target customers need to be screened accurately, wherein the target customers refer to customers who participate in the business activities at a high rate. However, due to the business experience limitations of business personnel (e.g., personnel engaged in the financial industry), target customers cannot be screened accurately, and the inventors have creatively proposed that, in response to a trigger operation to initiate a target business activity, target customers are screened from a full set of customers using a pre-trained assistant decision model. The triggering operation for starting the target business activity is optionally triggered by business personnel actively, and the full customer set refers to a set of customer data stored in a financial institution database.
In the embodiment of the present application, the auxiliary decision model is a decision tree model, and optionally, the auxiliary decision model is a gradient descent tree (GBDT) model, where the auxiliary decision model is formed by fusing a plurality of GBDT models. It should be noted that the assistant decision model of the embodiment of the present application selects the gradient descent tree model instead of the decision product of Experian, because the decision product of Experian mainly uses a single CART tree although providing the function of assisting in generating the decision rule. The CART tree is a binary tree, and the single CART tree model has the problems of insufficient fitting capability and poor robustness. Therefore, when the decision product of Experian is used, the relevant decision rule editing is mainly carried out depending on the abundant prior knowledge of business personnel.
In the embodiment of the application, the capturing rules of the target clients are successfully fitted in the pre-trained assistant decision model, so that the target clients can be accurately screened from the full client set by using the pre-trained assistant decision model. For example, for credit card credit line adjustment activities, the crawling rule that the assistant decision model fits to the target client may be: based on the capture rule, part of target clients meeting the conditions can be screened from the full client set.
S102, pushing target business activities to the screened target clients.
After the target customer is screened out through S101, the target business activity is pushed to the target customer, and optionally, the target business activity may be pushed to the target customer through a short message, a telephone call, or other manners.
In the embodiment of the application, in response to the triggering operation for developing the target business activity, the target clients can be screened from the full client set by directly utilizing the pre-trained assistant decision model, and then the target business activity is pushed to the target clients. Therefore, the target clients are screened by using the aid of the aid decision model, the condition that a service worker screens the target clients based on service experience is avoided, the service worker without the service experience can efficiently and accurately screen the target clients, and related activities are accurately pushed to clients with requirements.
Example two
Fig. 2 is a schematic flow chart of a method for assisting decision provided in the second embodiment of the present application, where the present embodiment is an optimization based on the foregoing embodiment, and adds a process of training an assistant decision model in advance, referring to fig. 2, the method includes:
s201, constructing an initial assistant decision model, training the initial assistant decision model sequentially by using a client sample, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models.
In the embodiment of the application, the initial assistant decision model is a gradient descent tree model; the super-parameters preset by the aid of the decision-making assisting model comprise the maximum depth of the gradient descent tree, the learning rate and the maximum number of the gradient descent tree. It should be noted that, in the process of constructing the initial aided decision-making model, the value range of each hyper-parameter of the preset initial aided decision-making model is obtained, so that in the process of training the initial training aided decision-making model, the value of each hyper-parameter is adjusted within the value range of each hyper-parameter, so as to obtain at least three target aided decision-making models.
Optionally, the hyper-parameter of the model may be adjusted based on a grid search, where the grid search may be a grid traversal search or a grid random search.
S202, selecting a preset number of target assistant decision models from the at least three target assistant decision models, and fusing the selected target assistant decision models to obtain a final assistant decision model.
In this embodiment, optionally, selecting a preset number of target assistant decision models from the at least three target assistant decision models includes: and evaluating the prediction effect of each target assistant decision model, and selecting a preset number of target assistant decision models according to the evaluation result.
The process of evaluating the predicted effect of any target assistant decision model comprises the following steps: and aiming at the target auxiliary decision model, calculating an AUC value of the target auxiliary decision model according to the predicted value and the actual value output by the target auxiliary decision model. The actual value is a value corresponding to the sample label, and the AUC value is used for evaluating the prediction effect of the target aid decision model. The AUC (area Under curve) value is the area enclosed by the receiver operating characteristic curve (ROC curve) and the coordinate axes. In the specific calculation, the ROC curve is determined by plotting a series of two different classification methods (boundary values or decision thresholds), the true positive rate is plotted as an ordinate and the false positive rate is plotted as an abscissa, and the false positive rate (the probability of being determined as a positive case but not a true case) and the true positive rate (the probability of being determined as a positive case but also a true case) are calculated from the predicted value and the actual value.
Further, in an alternative embodiment, the process of selecting a preset number of target assistant decision models according to the evaluation result includes: sequencing each target auxiliary decision model according to the AUC value of each target auxiliary decision model; and selecting a preset number of target assistant decision models according to the sorting result, wherein the preset number can be determined according to actual needs, and is exemplarily 3. Illustratively, the top N target decision-making aiding models with the largest AUC values are selected according to the descending order of the AUC values, or the target decision-making aiding models with AUC values larger than a preset threshold are selected.
And further fusing the selected target assistant decision model to obtain a final assistant decision model. In the embodiment of the present application, the final output of the assistant decision model is:
wherein X is each target assistance in the final aided decision modelAnd h is the predicted probability value of the client participating in the target business activity.
S203, responding to the trigger operation of starting the target business activity, and screening out the target client from the full-user set by using the final assistant decision model.
In an alternative embodiment, the customer data in the full customer set is sequentially input into the assistant decision model, and the target customer is determined according to the predicted value output by the assistant decision model.
S204, pushing the target business activity to the screened target client.
In the embodiment of the application, the auxiliary decision model is fused by a plurality of target auxiliary models with the best prediction effect, the robustness of the model is good, and when the trained auxiliary decision model is used for grabbing the target client, the accuracy rate of grabbing the target client can be improved, so that the target service activity can be pushed to the target client with the demand.
EXAMPLE III
Fig. 3 is a schematic flow chart of a method for assisting decision provided in the third embodiment of the present application, where this embodiment is a process of performing optimization and adding customer sample collection based on the foregoing embodiment, and referring to fig. 3, the method includes:
s301, obtaining an initial client sample set uploaded by service personnel, a target service type and a target service department to which a target service belongs.
The target business is exemplified by a credit card business, and the target business department is a credit card transaction department. It should be noted that the target service may also be another service, and is not specifically limited herein.
S302, judging whether the number of positive samples and the number of negative samples in the initial customer sample set reach a preset threshold value.
Wherein, the preset threshold value can be determined according to the actual training requirement. Illustratively, the preset threshold is 1000.
And S303, if not, extracting part of customer samples from samples generated by businesses similar to the target business type, and/or extracting part of customer samples from samples generated by other businesses of the target business department.
In the embodiment of the application, if the number of the samples in the initial customer sample set is determined to be insufficient, the customer samples of similar businesses and/or the customer samples of other businesses in the same department can be acquired for supplement. Thereby ensuring that a sufficient customer sample is obtained quickly.
And S304, supplementing the extracted part of the customer sample into the initial customer sample set.
Further, after supplementing the customer sample, the method further comprises:
judging whether the number of the samples in the supplemented initial client sample set is enough; and if not, supplementing the positive samples in the initial client sample set by adopting an oversampling mode, and extracting the negative samples to be supplemented from the global negative samples, wherein the global negative samples refer to the negative samples generated in the releasing process of all the existing services. Thus, with two replenishments, a sufficient customer sample can be obtained. And then carrying out a model training process according to S305-S308 to obtain a final assistant decision model.
It should be noted that the decision-making assistance system provides a front-end interface for obtaining samples, and if the business personnel does not obtain the initial customer sample set, the business personnel can check preset labels on the front-end interface or select similar sample supplementation.
S305, constructing an initial assistant decision model, training the sequentially initial training assistant decision model by utilizing the client samples in the initial client sample set, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models.
S306, selecting a preset number of target assistant decision models from the at least three target assistant decision models, and fusing the selected target assistant decision models to obtain a final assistant decision model.
For S305-S306, reference may be made to the description of the above embodiments, which are not repeated herein.
S307, responding to the trigger operation for starting the target business activity, and screening out the target customers from the full-scale customer set by using an auxiliary decision-making model.
S308, pushing the target business activity to the screened target client.
In the embodiment of the application, a sample obtaining method is provided, so that business personnel without business experience can obtain enough client samples through the method, and further normal operation of subsequent assistant decision model training is ensured.
Example four
Fig. 4 is a schematic flowchart of a method for assisting decision provided in embodiment 4 of the present application, where this embodiment is optimized based on the foregoing embodiments, and adds a process of acquiring and utilizing reflow data, see fig. 4, and the method includes:
s401, constructing an initial assistant decision model, training the initial assistant decision model sequentially by using a client sample, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models.
S402, selecting a preset number of target assistant decision models from the at least three target assistant decision models, and fusing the selected target assistant decision models to obtain a final assistant decision model.
And S403, responding to the trigger operation for starting the target business activity, and screening out the target customers from the full-scale customer set by using a pre-trained auxiliary decision model.
S404, pushing target business activities to the screened target clients.
In the prior art, data does not form a complete closed loop, namely, the result of previous decision cannot be fed back and the capturing rule of the previous target client cannot be calibrated. Based on the above, the inventor creatively proposes to acquire the backflow data and optimize the assistant decision model by using the backflow data. See, in particular, S405-S407.
S405, obtaining result backflow data.
The result reflow data comprises client data of actually participating and not participating in the target business activity, and the client data at least comprises the time of the client participating in the target business activity. Since the reflow data is stored in a database table, in an alternative embodiment, reflow data for a specified length of time (e.g., the last N days) may be periodically retrieved from the database table.
And S406, taking the client data actually participating in the target business activity as a positive sample, and taking the client data not participating in the target business activity as a negative sample.
And S407, correcting the training aid decision model by using the positive sample and the negative sample.
That is, the decision-aiding model continues to be trained by using the positive and negative samples determined in S406. For a specific training process, reference is made to the above embodiments, which are not described herein again.
And S408, replacing the original assistant decision model with the assistant decision model after the correction training in a gray scale publishing mode.
In the embodiment of the application, the assistant decision model after retraining is deployed in a gray scale publishing manner, so that the situation that the prediction accuracy of the retrained assistant decision model is lower than that of the original assistant decision model is avoided.
In the embodiment of the application, the reflux data are obtained, and the assistant decision model is retrained by utilizing the reflux data, so that the whole data flow can form a closed loop, the assistant training model can be continuously optimized along with the proceeding of business activities, and a better prediction effect is obtained.
EXAMPLE five
Fig. 5 is a schematic structural diagram of a system for assisting decision in a fifth embodiment of the present application, where the system is used in a financial scenario to accurately screen a target customer and push a target business activity to the target customer, and referring to fig. 5, the system includes:
the model training module 501 is used for constructing an initial assistant decision model, sequentially training the initial assistant decision model by using a client sample, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models;
a fusion module 502, configured to select a preset number of target assistant decision models from the at least three target assistant decision models, and fuse the selected target assistant decision models to obtain a final assistant decision model;
an assistant decision module 503, configured to, in response to a trigger operation for starting a target business activity, screen out a target client from a full-scale client set by using an assistant decision model;
and a pushing module 504, configured to push the target business activity to the screened target client.
In the embodiment of the application, in response to the triggering operation for starting the target business activity, the trained assistant decision model is utilized to screen out the target clients from the full amount of client sets, and then the target business activity is pushed to the target clients. Therefore, the target clients are screened by using the aid of the aid decision model, the condition that a service worker screens the target clients based on service experience is avoided, the service worker without the service experience can efficiently and accurately screen the target clients, and related activities are accurately pushed to clients with requirements.
On the basis of the above implementation, optionally, the fusion module includes:
and the screening unit is used for evaluating the prediction effect of each target auxiliary decision model and selecting a preset number of target auxiliary decision models according to the evaluation result.
On the basis of the above implementation, optionally, the final output of the assistant decision model is:
wherein X is the mean value of the predicted values of all target auxiliary decision models in the final auxiliary decision model, and h is the predicted probability value of the client participating in the target business activity;
correspondingly, the assistant decision module is further configured to:
and sequentially inputting the customer data in the full customer set into the assistant decision-making model, and determining the target customer according to the predicted value output by the assistant decision-making model.
On the basis of the implementation, optionally, the auxiliary decision model is a gradient descent tree model;
correspondingly, the super-parameters preset by the aid of the decision-making assisting model comprise the maximum depth of the gradient descent tree, the learning rate and the maximum number of the gradient descent tree.
On the basis of the above implementation, optionally, the training unit further includes an evaluation subunit, where the evaluation subunit is configured to:
and aiming at any target auxiliary decision model, calculating an AUC (AUC) value of the target auxiliary decision model according to the predicted value and the actual value output by the target auxiliary decision model, wherein the AUC value is used for evaluating the prediction effect of the target auxiliary decision model.
On the basis of the above implementation, optionally, the fusion unit further includes a screening subunit, and the screening subunit is configured to:
sequencing each target auxiliary decision model according to the AUC value of each target auxiliary decision model;
and selecting a preset number of target assistant decision models from the sorting results.
On the basis of the above implementation, optionally, the system further includes:
and the sample acquisition module is used for acquiring an initial client sample set, a target service type and a target service department to which the target service belongs, which are uploaded by service personnel, before the assistant decision-making model is trained.
On the basis of the above implementation, optionally, the system further includes:
the first sample judging module is used for judging whether the quantity of positive samples and the quantity of negative samples in the initial client sample set reach a preset threshold value or not;
the sample extraction module is used for extracting part of customer samples from samples generated by businesses similar to the target business type and/or extracting part of customer samples from samples generated by other businesses of the target business department if the judgment result is negative;
and the first sample supplementing module is used for supplementing the extracted part of the customer sample into the initial customer sample set.
On the basis of the above implementation, optionally, the system further includes:
the second judgment module is used for judging whether the number of samples in the supplemented initial customer sample set is enough or not after the customer samples are supplemented;
and the second sample supplementing module is used for supplementing the positive samples in the initial client sample set in an oversampling mode and extracting the negative samples to be supplemented from the global negative samples if the judgment result is negative, wherein the global negative samples refer to the negative samples generated in the process of putting all the existing services.
On the basis of the above implementation, optionally, the system further includes:
and the backflow data acquisition module is used for acquiring result backflow data after target business activities are pushed to the screened target clients, wherein the result backflow data comprise client data which actually participate in the target business activities and client data which do not participate in the target business activities.
On the basis of the above implementation, optionally, the system further includes a repetitive training module, configured to:
taking the client data actually participating in the target business activity as a positive sample, and taking the client data not participating in the target business activity as a negative sample;
and correcting the training aid decision model by using the positive sample and the negative sample.
On the basis of the above implementation, optionally, the system further includes:
and the issuing module is used for replacing the corrected and trained assistant decision model with the original assistant decision model in a gray level issuing mode after the correction and training assistant decision model is corrected.
The system for assisting decision provided by the embodiment of the application can execute the method for assisting decision provided by any embodiment of the application, and has corresponding functional modules and beneficial effects of the execution method.
EXAMPLE six
Fig. 6 is a schematic structural diagram of a server according to a sixth implementation of the present application. FIG. 6 illustrates a block diagram of an exemplary server 12 suitable for use in implementing embodiments of the present application. The server 12 shown in fig. 6 is only an example, and should not bring any limitation to the function and the scope of use of the embodiments of the present application.
As shown in FIG. 6, the server 12 is in the form of a general purpose computing device. The components of the server 12 may include, but are not limited to: one or more processors or processing units 16, a system memory 28, and a bus 18 that couples various system components including the system memory 28 and the processing unit 16.
The server 12 typically includes a variety of computer system readable media. Such media may be any available media that is accessible by server 12 and includes both volatile and nonvolatile media, removable and non-removable media.
The system memory 28 may include computer system readable media in the form of volatile memory, such as Random Access Memory (RAM)30 and/or cache memory 32. The server 12 may further include other removable/non-removable, volatile/nonvolatile computer system storage media. By way of example only, storage system 34 may be used to read from and write to non-removable, nonvolatile magnetic media (not shown in FIG. 6, and commonly referred to as a "hard drive"). Although not shown in FIG. 6, a magnetic disk drive for reading from and writing to a removable, nonvolatile magnetic disk (e.g., a "floppy disk") and an optical disk drive for reading from or writing to a removable, nonvolatile optical disk (e.g., a CD-ROM, DVD-ROM, or other optical media) may be provided. In these cases, each drive may be connected to bus 18 by one or more data media interfaces. System memory 28 may include at least one program product having a set (e.g., at least one) of program modules that are configured to carry out the functions of embodiments of the application.
A program/utility 40 having a set (at least one) of program modules 42 may be stored, for example, in system memory 28, such program modules 42 including, but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which examples or some combination thereof may comprise an implementation of a network environment. Program modules 42 generally perform the functions and/or methodologies of the embodiments described herein.
The server 12 may also communicate with one or more external devices 14 (e.g., keyboard, pointing device, etc.) or a display 24, may also communicate with one or more devices that enable a user to interact with the server 12, and/or may communicate with any device (e.g., network card, modem, etc.) that enables the server 12 to communicate with one or more other computing devices. Such communication may be through an input/output (I/O) interface 22. Also, the server 12 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network, such as the Internet) via the network adapter 20. As shown, the network adapter 20 communicates with the other modules of the server 12 via the bus 18. It should be understood that although not shown in the figures, other hardware and/or software modules may be used in conjunction with the server 12, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, among others.
The processing unit 16 executes programs stored in the system memory 28 to execute various functional applications and data processing, such as a method for assisting decision making provided by the embodiments of the present application, the method including:
constructing an initial assistant decision model, training the initial assistant decision model in sequence by utilizing a client sample, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models;
selecting a preset number of target assistant decision models from the at least three target assistant decision models, and fusing the selected target assistant decision models to obtain a final assistant decision model;
responding to a trigger operation for starting a target business activity, and screening out a target client from a full client set by using an auxiliary decision model;
and pushing the target business activity to the screened target client.
EXAMPLE seven
A storage medium, in particular a computer-readable storage medium, is further provided in an embodiment of the present application, where a computer program is stored on the storage medium, and when the computer program is executed by a processor, the method for assisting decision making provided in the embodiment of the present application is implemented, where the method includes:
constructing an initial assistant decision model, training the initial assistant decision model in sequence by utilizing a client sample, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models;
selecting a preset number of target assistant decision models from the at least three target assistant decision models, and fusing the selected target assistant decision models to obtain a final assistant decision model;
responding to a trigger operation for starting a target business activity, and screening out a target client from a full client set by using an auxiliary decision model;
and pushing the target business activity to the screened target client.
The storage media of the embodiments of the present application may take any combination of one or more computer-readable media. The computer readable medium may be a computer readable signal medium or a computer readable storage medium. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
A computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present application and the technical principles employed. It will be understood by those skilled in the art that the present application is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the application. Therefore, although the present application has been described in more detail with reference to the above embodiments, the present application is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present application, and the scope of the present application is determined by the scope of the appended claims.
Claims (15)
1. A method for aiding decision making, the method comprising:
constructing an initial assistant decision model, training the initial assistant decision model in sequence by utilizing a client sample, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models;
selecting a preset number of target assistant decision models from the at least three target assistant decision models, and fusing the selected target assistant decision models to obtain a final assistant decision model;
responding to a trigger operation for starting a target business activity, and screening out a target client from a full-scale client set by using the assistant decision model;
and pushing the target business activity to the screened target client.
2. The method of claim 1, wherein selecting a preset number of target aided decision models from the at least three target aided decision models comprises:
and evaluating the prediction effect of each target assistant decision model, and selecting a preset number of target assistant decision models according to the evaluation result.
3. The method of claim 1, wherein the final aided decision model output is:
wherein X is the mean value of the predicted values of all target auxiliary decision models in the final auxiliary decision model, and h is the predicted probability value of the client participating in the target business activity;
correspondingly, the method for screening out the target customers from the full-scale customer set by using the assistant decision model comprises the following steps:
and sequentially inputting the customer data in the full customer set into the assistant decision-making model, and determining the target customer according to the predicted value output by the assistant decision-making model.
4. The method according to any of claims 1-3, wherein the aided decision model is a gradient descent tree model;
correspondingly, the super-parameters preset by the aid of the decision-making assisting model comprise the maximum depth of the gradient descent tree, the learning rate and the maximum number of the gradient descent tree.
5. The method of claim 2, wherein the step of evaluating the predicted effect of each objective aid decision model comprises:
and aiming at any target auxiliary decision model, calculating an AUC (AUC) value of the target auxiliary decision model according to a predicted value and an actual value output by the target auxiliary decision model, wherein the AUC value is used for evaluating the prediction effect of the target auxiliary decision model.
6. The method of claim 5, wherein the step of selecting a predetermined number of target-aided decision-making models according to the evaluation result comprises:
sequencing each target auxiliary decision model according to the AUC value of each target auxiliary decision model;
and selecting a preset number of target assistant decision models from the sorting results.
7. The method of claim 1, wherein prior to training the aided decision model, the method further comprises:
and acquiring an initial client sample set, a target service type and a target service department to which the target service belongs, which are uploaded by service personnel.
8. The method of claim 7, further comprising:
judging whether the number of positive samples and the number of negative samples in the initial client sample set reach a preset threshold value or not;
if not, extracting part of customer samples from samples generated by businesses similar to the target business type, and/or extracting part of customer samples from samples generated by other businesses of the target business department;
and supplementing the extracted part of the customer sample into the initial customer sample set.
9. The method of claim 8, wherein after replenishing the customer sample, the method further comprises:
judging whether the number of the samples in the supplemented initial client sample set is enough;
and if not, supplementing the positive samples in the initial client sample set by adopting an oversampling mode, and extracting the negative samples to be supplemented from the global negative samples, wherein the global negative samples refer to the negative samples generated in the releasing process of all the existing services.
10. The method of claim 1, wherein after pushing the targeted business activity to the screened targeted customer, the method further comprises:
and acquiring result reflow data, wherein the result reflow data comprises client data which actually participate in and do not participate in the target business activities.
11. The method of claim 10, further comprising:
taking the client data actually participating in the target business activity as a positive sample, and taking the client data not participating in the target business activity as a negative sample;
and correcting and training an auxiliary decision model by using the positive sample and the negative sample.
12. The method of claim 11, wherein after revising the trained aided decision model, the method further comprises:
and replacing the original assistant decision model with the assistant decision model after the correction training in a gray scale publishing mode.
13. A system for aiding decision making, the system comprising:
the model training module is used for constructing an initial assistant decision model, sequentially training the initial assistant decision model by utilizing client samples, and adjusting the hyper-parameters of the initial assistant decision model in the training process to obtain at least three target assistant decision models;
the fusion module is used for selecting a preset number of target auxiliary decision models from the at least three target auxiliary decision models and fusing the selected target auxiliary decision models to obtain a final auxiliary decision model;
the assistant decision-making module is used for responding to the triggering operation of starting the target business activity and screening out target customers from the full-scale customer set by utilizing the assistant decision-making module;
and the pushing module is used for pushing the target business activity to the screened target client.
14. A server, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement a method of aiding decision making as claimed in any one of claims 1-12.
15. A storage medium on which a computer program is stored which, when being executed by a processor, carries out a method for assisting decision making according to any one of claims 1-12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011025802.XA CN112184304A (en) | 2020-09-25 | 2020-09-25 | Method, system, server and storage medium for assisting decision |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011025802.XA CN112184304A (en) | 2020-09-25 | 2020-09-25 | Method, system, server and storage medium for assisting decision |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112184304A true CN112184304A (en) | 2021-01-05 |
Family
ID=73944246
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011025802.XA Pending CN112184304A (en) | 2020-09-25 | 2020-09-25 | Method, system, server and storage medium for assisting decision |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112184304A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113191812A (en) * | 2021-05-12 | 2021-07-30 | 深圳索信达数据技术有限公司 | Service recommendation method, computer device and computer-readable storage medium |
| CN113837866A (en) * | 2021-09-29 | 2021-12-24 | 重庆富民银行股份有限公司 | Two-stage management method and system based on full stock customer |
| CN114143188A (en) * | 2021-11-18 | 2022-03-04 | 深圳大学 | Video processing method and device, electronic equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109034660A (en) * | 2018-08-22 | 2018-12-18 | 平安科技（深圳）有限公司 | Based on the determination method and relevant apparatus of the risk control strategy of prediction model |
| CN109117864A (en) * | 2018-07-13 | 2019-01-01 | 华南理工大学 | Coronary heart disease risk prediction technique, model and system based on heterogeneous characteristic fusion |
| CN109214587A (en) * | 2018-09-27 | 2019-01-15 | 重庆智万家科技有限公司 | A kind of demand for commodity prediction based on three decisions divides storehouse planing method with logistics |
| WO2019061187A1 (en) * | 2017-09-28 | 2019-04-04 | 深圳乐信软件技术有限公司 | Credit evaluation method and apparatus and gradient boosting decision tree parameter adjustment method and apparatus |
| CN110659922A (en) * | 2018-06-28 | 2020-01-07 | 马上消费金融股份有限公司 | Client screening method, device, server and computer readable storage medium |
| CN110991476A (en) * | 2019-10-18 | 2020-04-10 | 北京奇艺世纪科技有限公司 | Training method and device for decision classifier, recommendation method and device for audio and video, and storage medium |
| CN111401766A (en) * | 2020-03-24 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Model, service processing method, device and equipment |
-
2020
- 2020-09-25 CN CN202011025802.XA patent/CN112184304A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019061187A1 (en) * | 2017-09-28 | 2019-04-04 | 深圳乐信软件技术有限公司 | Credit evaluation method and apparatus and gradient boosting decision tree parameter adjustment method and apparatus |
| CN110659922A (en) * | 2018-06-28 | 2020-01-07 | 马上消费金融股份有限公司 | Client screening method, device, server and computer readable storage medium |
| CN109117864A (en) * | 2018-07-13 | 2019-01-01 | 华南理工大学 | Coronary heart disease risk prediction technique, model and system based on heterogeneous characteristic fusion |
| CN109034660A (en) * | 2018-08-22 | 2018-12-18 | 平安科技（深圳）有限公司 | Based on the determination method and relevant apparatus of the risk control strategy of prediction model |
| CN109214587A (en) * | 2018-09-27 | 2019-01-15 | 重庆智万家科技有限公司 | A kind of demand for commodity prediction based on three decisions divides storehouse planing method with logistics |
| CN110991476A (en) * | 2019-10-18 | 2020-04-10 | 北京奇艺世纪科技有限公司 | Training method and device for decision classifier, recommendation method and device for audio and video, and storage medium |
| CN111401766A (en) * | 2020-03-24 | 2020-07-10 | 支付宝(杭州)信息技术有限公司 | Model, service processing method, device and equipment |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113191812A (en) * | 2021-05-12 | 2021-07-30 | 深圳索信达数据技术有限公司 | Service recommendation method, computer device and computer-readable storage medium |
| CN113191812B (en) * | 2021-05-12 | 2024-02-02 | 深圳索信达数据技术有限公司 | Service recommendation method, computer equipment and computer readable storage medium |
| CN113837866A (en) * | 2021-09-29 | 2021-12-24 | 重庆富民银行股份有限公司 | Two-stage management method and system based on full stock customer |
| CN114143188A (en) * | 2021-11-18 | 2022-03-04 | 深圳大学 | Video processing method and device, electronic equipment and storage medium |
| CN114143188B (en) * | 2021-11-18 | 2023-06-23 | 深圳大学 | Video processing method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190325514A1 (en) | Credit risk prediction method and device based on lstm model | |
| CN108171260B (en) | Picture identification method and system | |
| CN112184304A (en) | Method, system, server and storage medium for assisting decision | |
| CN109961165B (en) | Method, device, equipment and storage medium for predicting part quantity | |
| CN112990294B (en) | Training method and device of behavior discrimination model, electronic equipment and storage medium | |
| CN111783645A (en) | Character recognition method and device, electronic equipment and computer readable storage medium | |
| CN111210335A (en) | User risk identification method and device and electronic equipment | |
| CN112613569A (en) | Image recognition method, and training method and device of image classification model | |
| CN111210336A (en) | User risk model generation method and device and electronic equipment | |
| CN113947336A (en) | Method, device, storage medium and computer equipment for evaluating risks of bidding enterprises | |
| CN113034046A (en) | Data risk metering method and device, electronic equipment and storage medium | |
| CN111210332A (en) | Method and device for generating post-loan management strategy and electronic equipment | |
| CN111815435A (en) | Visualization method, device, equipment and storage medium for group risk characteristics | |
| CN114090601B (en) | Data screening method, device, equipment and storage medium | |
| CN111046184A (en) | Text risk identification method, device, server and storage medium | |
| CN113051911B (en) | Method, apparatus, device, medium and program product for extracting sensitive words | |
| CN114066603A (en) | Post-loan risk early warning method and device, electronic equipment and computer readable medium | |
| CN112950392A (en) | Information display method, posterior information determination method and device and related equipment | |
| CN114565030B (en) | Feature screening method and device, electronic equipment and storage medium | |
| CN111681115B (en) | Payday identification method, system, equipment and storage medium | |
| EP4089592A1 (en) | Method for determining annotation capability information, related apparatus and computer program product | |
| WO2022066034A1 (en) | Automating debt collection processes using artificial intelligence | |
| CN117037167A (en) | Sensitive information detection method, device, equipment and medium based on artificial intelligence | |
| CN117436454A (en) | Data identification method, device, electronic equipment and storage medium | |
| CN117094391A (en) | Causal event pair processing method, causal event pair processing device, causal event pair processing computer equipment and storage medium |
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 |