CN115660722A

CN115660722A - Prediction method and device for silver life customer conversion and electronic equipment

Info

Publication number: CN115660722A
Application number: CN202211670420.1A
Authority: CN
Inventors: 黄业宇
Original assignee: Ping An Bank Co Ltd
Current assignee: Ping An Bank Co Ltd
Priority date: 2022-12-26
Filing date: 2022-12-26
Publication date: 2023-01-31
Anticipated expiration: 2042-12-26
Also published as: CN115660722B

Abstract

The invention provides a prediction method, a device and electronic equipment for silver-life customer conversion, which belong to the technical field of data processing, wherein the intimacy between each nonlinear customer and a current bank, the intimacy correlation factor of each nonlinear customer, the basic attribute of each customer and the correlation attribute of a customer agent of each customer are input into a trained silver-life customer conversion prediction model for silver-life customer conversion prediction.

Description

Prediction method and device for silver life customer conversion and electronic equipment

Technical Field

The invention relates to the technical field of data processing, in particular to a prediction method and device for silver-life customer conversion and electronic equipment.

Background

With the arrival of the big data era and the development of financial technologies, banks or insurance companies have developed own technology departments, and a large amount of business data and data mining models are produced, but nevertheless, the following problems still exist: the financial industry has a natural information barrier. Due to the problems of financial supervision and industry competition, financial companies do not cooperate with internet enterprises much, and are usually only concentrated in the fields of the financial companies and information is blocked.

At present, in the financial industry, a traditional silver-life customer transformation model system is usually realized by adopting expert experience rules, and the expert experience rules have the characteristics of strong interpretability, high controllability and full fitting with business historical data, but the expert experience rules are subjective, so the accuracy of the predicted silver-life customer transformation probability is poor; besides expert experience rules, a logistic model is generally used for predicting customer conversion of the traditional financial enterprise, the logistic model has good model interpretability, but the logistic model is a linear model in nature, and the probability accuracy of the predicted conversion of the silver-life customer is poor; an xgboost model is used for conversion prediction of a client, and under the condition of small data volume and few feature labels, the prediction accuracy is relatively high, but in the xgboost modeling process, all modeling data need to be loaded into a memory for pre-sequencing, and not only a feature value but also a sequencing index of features need to be stored, which inevitably causes too large memory consumption, and when the xgboost model is applied to a client conversion prediction scene with large data and multiple feature labels, the prediction accuracy is poor, that is, the requirement of service accuracy cannot be well met.

In conclusion, the existing prediction method for the silver life customer conversion has the technical problem of poor accuracy.

Disclosure of Invention

In view of the above, the present invention provides a method, an apparatus, and an electronic device for predicting a conversion of a silver-life customer, so as to alleviate the technical problem of poor accuracy of the existing method for predicting a conversion of a silver-life customer.

In a first aspect, an embodiment of the present invention provides a method for predicting a conversion of a silver-based customer, including:

obtaining client service data in a historical time range, and determining initial values of various characteristics of each client according to the client service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment dendrogram, and the intimacy assessment dendrogram comprises the following steps from top to bottom: a label layer, a criterion layer, a sub-criterion layer and the index layer;

performing affinity evaluation based on the affinity evaluation dendrogram and initial values of various features of each customer to obtain affinity of each customer with a current bank and affinity correlation factors of each customer, wherein the affinity correlation factors are scores of various features contained in the label layer, the criterion layer, the sub-criterion layer and the index layer in the affinity evaluation dendrogram;

determining basic attributes of each client and relevant attributes of a client agent of each client according to the client service data;

and inputting the intimacy of each customer and the current bank, the intimacy correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into a trained silver-life customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver-life customer conversion prediction model is a lightGBM model.

In a second aspect, an embodiment of the present invention further provides an apparatus for predicting a conversion of a silver-life customer, including:

the system comprises an acquisition and determination unit, a data processing unit and a data processing unit, wherein the acquisition and determination unit is used for acquiring customer service data in a historical time range and determining initial values of various characteristics of each customer according to the customer service data, the various characteristics are characteristics of an index layer in an intimacy assessment tree-like graph, and the intimacy assessment tree-like graph comprises the following steps from top to bottom: a label layer, a criterion layer, a sub-criterion layer and the index layer;

the intimacy evaluation unit is used for carrying out intimacy evaluation on the basis of the intimacy evaluation tree diagram and initial values of various characteristics of each client to obtain intimacy of each client and a current bank and intimacy related factors of each client, wherein the intimacy related factors are scores of various characteristics contained in the label layer, the criterion layer, the sub-criterion layer and the index layer in the intimacy evaluation tree diagram;

a determining unit, configured to determine a basic attribute of each client and a related attribute of a client agent of each client according to the client service data;

and the silver-life customer conversion prediction unit is used for inputting the intimacy degree of each customer and the current bank, the intimacy degree correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into a trained silver-life customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver-life customer conversion prediction model is a lightGBM model.

In a third aspect, an embodiment of the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor executes the computer program to implement the steps of the method according to any one of the first aspect.

In a fourth aspect, embodiments of the present invention also provide a computer-readable storage medium storing machine-executable instructions that, when invoked and executed by a processor, cause the processor to perform the method of any of the first aspects.

In an embodiment of the present invention, a method for predicting a silver-life customer conversion is provided, including: acquiring customer service data in a historical time range, and determining initial values of various characteristics of each customer according to the customer service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment tree diagram, and the intimacy assessment tree diagram comprises the following steps from top to bottom: the system comprises a label layer, a criterion layer, a sub-criterion layer and an index layer; performing affinity evaluation based on the affinity evaluation dendrogram and initial values of various characteristics of each client to obtain the affinity of each client and the current bank and affinity correlation factors of each client, wherein the affinity correlation factors are scores of various characteristics contained in a label layer, a criterion layer, a sub-criterion layer and an index layer in the affinity evaluation dendrogram; determining basic attributes of each client and relevant attributes of a client agent of each client according to the client service data; and inputting the intimacy of each customer and the current bank, the intimacy correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into the trained silver-life customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver-life customer conversion prediction model is a lightGBM model. According to the prediction method for the silver-life customer conversion, the nonlinear intimacy between each customer and the current bank and the parameters of the intimacy correlation factor of each customer are obtained based on the customer service data, and then the intimacy between each customer and the current bank, the intimacy correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer are input into the trained silver-life customer conversion prediction model to perform the prediction of the silver-life customer conversion.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.

FIG. 1 is a flow chart of a method for predicting the conversion of a silver-based customer according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of an affinity assessment tree according to an embodiment of the present invention;

FIG. 3 is a thermodynamic diagram of a customer in service penetration, interaction activity, quality experience, and negative impact on several dimensions provided by an embodiment of the present invention;

FIG. 4 is a schematic diagram illustrating relationships between affinity and contract number, number of types of products held, VIP coverage, etc. according to an embodiment of the present invention;

FIG. 5 is a schematic diagram illustrating the relationship between affinity and AUM provided by the embodiment of the present invention;

FIG. 6 is a schematic diagram of an apparatus for predicting customer conversion of silver life according to an embodiment of the present invention;

fig. 7 is a schematic diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

The technical solutions of the present invention will be described clearly and completely with reference to the following embodiments, and it should be apparent that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.

The traditional prediction method for the silver life customer conversion is poor in accuracy.

Based on the above, the method for predicting silver life customer conversion obtains parameters of the non-linear intimacy degree between each customer and the current bank and the intimacy degree correlation factor of each customer based on the customer service data, and then inputs the intimacy degree between each customer and the current bank, the intimacy degree correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into the trained silver life customer conversion prediction model for predicting silver life customer conversion.

To facilitate understanding of the present embodiment, a method for predicting the conversion of the silver-based customer disclosed in the present embodiment will be described in detail.

The first embodiment is as follows:

in accordance with an embodiment of the present invention, there is provided an embodiment of a method for predicting a conversion of a silver-life customer, it is noted that the steps illustrated in the flowchart of the accompanying drawings may be performed in a computer system such as a set of computer-executable instructions, and that while a logical order is illustrated in the flowchart, in some cases the steps illustrated or described may be performed in an order different than here.

Fig. 1 is a flowchart of a method for predicting the conversion of a silver-based customer according to an embodiment of the present invention, as shown in fig. 1, the method includes the following steps:

step S102, obtaining customer service data in a historical time range, and determining initial values of various characteristics of each customer according to the customer service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment dendrogram, and the intimacy assessment dendrogram comprises the following steps from top to bottom: the system comprises a label layer, a criterion layer, a sub-criterion layer and an index layer;

in the embodiment of the present invention, the historical time range may be a past one month time, the customer service data is obtained from a financial institution, the financial institution is a department of the current bank, and has all service data of professional subsidiaries of different sizes of the whole current bank group, the customer service data in the past one month time generally has customer service data of more than 2 hundred million customers, and the data volume is very huge.

After the customer service data is obtained, initial values of various characteristics of each customer can be determined according to the customer service data, the various characteristics are characteristics of an index layer in an intimacy assessment tree map, and fig. 2 shows the intimacy assessment tree map, wherein a layer marked with 1 is a label layer, a layer marked with 2 is a criterion layer, a layer marked with 3 is a sub-criterion layer, and a layer marked with 4 is an index layer.

Specifically, after the client service data is obtained, initial values of various characteristics of each client can be processed and obtained, for example, according to data of APP activity in the client service data, the initial values of APP activity of each client are processed and obtained, for example, the number of times of the client a uses APP r is 18, the number of times of the client a uses APP s is 12, and the number of times of the client t is 30, then the initial value of the APP activity of the client a may be 18+12+30=60, and similarly, the initial values of the APP activity of the client B, the initial values of the client C, and the like may be obtained, which is only a simple example, and the initial values of various characteristics may also be determined in other manners, and are not particularly limited herein.

Step S104, performing affinity evaluation based on the affinity evaluation dendrogram and initial values of various characteristics of each client to obtain the affinity of each client and a current bank and affinity correlation factors of each client, wherein the affinity correlation factors are scores of various characteristics contained in a label layer, a criterion layer, a sub-criterion layer and an index layer in the affinity evaluation dendrogram;

the inventor considers that the existing customer characteristics are usually concentrated on the characteristics of linear dimensions such as basic attributes, transaction data and financial data, and the linear characteristics sometimes can not well describe the portrait of the customer, so the inventor constructs the nonlinear and hierarchical characteristics in the invention to better describe the portrait attributes of the customer, namely, the intimacy assessment of the invention realizes the breakthrough of data analysis technology, fully mines the service value in the data, and combines the advantages of big data to achieve the effect of enabling the service of the data.

As can be seen from the intimacy degree assessment tree diagram in fig. 2, the label layer, the criterion layer, the sub-criterion layer, and the index layer have a very large number of features, that is, many dimensions of data, and the portrait of the customer can be described through the multidimensional data features (that is, the intimacy degree between the customer and the current bank and the intimacy degree related factor of the customer), so that the features of the customer can be described more precisely, accurately, and scientifically, and the conversion probability of the customer obtained based on the features is more accurate.

The process is described in detail below and will not be described herein.

Step S106, determining the basic attribute of each client and the related attribute of the client agent of each client according to the client service data;

specifically, the basic attribute of each client refers to the conventional existing client characteristics, such as basic information, transaction data, financial data, and the like, and the related attribute of the client agent of each client refers to the important client number information of the client agent (e.g., client manager) of each client, whether the client agent of each client is a star agent, and the like.

And step S108, inputting the intimacy between each customer and the current bank, the intimacy correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into the trained silver-life customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver-life customer conversion prediction model is a lightGBM model.

Therefore, compared with the traditional scheme, the method has the advantages that the customer characteristics are richer and more scientific in the scheme, the intimacy between each customer and the current bank and the intimacy correlation factor between each customer are increased on the basis of the original customer characteristics, in addition, the scheme of the invention also replaces a prediction model, the silver-life customer conversion prediction model of the invention is a lightGBM (light gradient raiser) model, the lightGBM (light gradient raiser) model has the characteristics of interpretability and high-efficiency prediction, and the lightGBM is a data scientific tool which is rapid and efficient, low in memory occupation, high in accuracy and capable of supporting parallel and large-scale data processing. Therefore, the model is suitable for a big data multi-feature prediction scene, the accuracy of the prediction result in the scene is good, and the time consumption is short.

In an embodiment of the present invention, a method for predicting a conversion of a silver-life customer is provided, including: acquiring customer service data in a historical time range, and determining initial values of various characteristics of each customer according to the customer service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment tree diagram, and the intimacy assessment tree diagram comprises the following steps from top to bottom: the system comprises a label layer, a criterion layer, a sub-criterion layer and an index layer; performing affinity evaluation based on the affinity evaluation dendrogram and initial values of various characteristics of each client to obtain the affinity of each client and the current bank and affinity correlation factors of each client, wherein the affinity correlation factors are scores of various characteristics contained in a label layer, a criterion layer, a sub-criterion layer and an index layer in the affinity evaluation dendrogram; determining basic attributes of each client and relevant attributes of a client agent of each client according to the client service data; and inputting the intimacy of each customer and the current bank, the intimacy correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into the trained silver-life customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver-life customer conversion prediction model is a lightGBM model. According to the prediction method for the silver-life customer conversion, the nonlinear intimacy between each customer and the current bank and the parameters of the intimacy correlation factor of each customer are obtained based on the customer service data, and then the intimacy between each customer and the current bank, the intimacy correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer are input into the trained silver-life customer conversion prediction model to perform the prediction of the silver-life customer conversion.

The foregoing briefly describes the method for predicting the conversion of the silver-life customer according to the present invention, and the details thereof are described in detail below.

In an optional embodiment of the present invention, the step S104 of performing affinity assessment based on the affinity assessment dendrogram and initial values of various features of each client specifically includes the following steps:

(1) Acquiring first characteristic value intervals corresponding to various characteristics of the index layer and weights corresponding to the first characteristic value intervals;

to facilitate an understanding of this process, the following is illustrated as an example:

for example, the characteristics of the APP activity and the 95511 activity in the index layer have first eigenvalue intervals of [0, 30], (30, 60], (60, + ∞) respectively corresponding to the characteristics of the APP activity, the weights of the first eigenvalue intervals are 0.3, 0.5 and 0.2 respectively, the first eigenvalue intervals of [0, 50], (50, + ∞) respectively corresponding to the characteristics of the 95511 activity, and the weights of the first eigenvalue intervals are 0.6 and 0.4 respectively.

(2) Determining a target first characteristic value interval to which initial values of various characteristics of the same client downstream of the first target characteristic of the sub-criterion layer belong, and determining a weight corresponding to the target first characteristic value interval, wherein the first target characteristic is any one of the characteristics of the sub-criterion layer;

for example, the first target feature is channel interaction strength, and various features downstream thereof include: APP activity and 95511 activity. If the initial value of APP activity of one client a is 40 and the initial value of 95511 activity is 60, the target first characteristic value interval to which the initial value of APP activity 40 of the client a downstream of the channel interaction strength is determined belongs is (30, 60), and the corresponding weight is 0.5 (see the example in (1) above), and the target first characteristic value interval to which the initial value of 95511 activity 60 of the client a belongs is (50, + ∞), and the corresponding weight is 0.4.

(3) Weighting and calculating initial values of various characteristics of the same client at the downstream of the first target characteristic of the sub-criterion layer and weights corresponding to target first characteristic value intervals to which the initial values belong to obtain scores of the first target characteristics of the clients;

as shown in the above example (2), weighting calculation is performed on the initial value 40 of the APP activity of the client a downstream of the channel interaction strength and the weight 0.5 corresponding to the target first characteristic value interval (30, 60) to which 40 belongs, and the initial value 60 of the 95511 activity of the client a and the weight 0.4 corresponding to the target first characteristic value interval (50, + ∞) to which 60 belongs, so as to obtain 40 × 0.5+60 × 0.4=44, which is the score of the channel interaction strength of the client a.

Similarly, scores may be obtained for various first target features (i.e., each feature in the sub-criteria layer) for each customer.

(4) Acquiring second characteristic value intervals corresponding to various first target characteristics of the sub-criterion layer and weights corresponding to the second characteristic value intervals;

(5) Determining a target second characteristic value interval to which the scores of various first target characteristics of the same client downstream of a second target characteristic of the criterion layer belong, and determining the weight corresponding to the target second characteristic value interval, wherein the second target characteristic is any one of the characteristics of the criterion layer;

(6) Weighting and calculating the scores of various first target characteristics of the same client at the downstream of the second target characteristics of the criterion layer and the weights corresponding to the target second characteristic value intervals to which the scores belong to obtain the scores of the second target characteristics of each client;

the processes of (4) to (6) described above are similar to the processes of (1) to (3) described above, and a description thereof will not be repeated.

(7) Acquiring third characteristic value intervals corresponding to various second target characteristics of the criterion layer and weights corresponding to the third characteristic value intervals;

(8) Determining a target third characteristic value interval to which the scores of various second target characteristics of the same customer downstream of the characteristics of the intimacy degree of the label layer belong, and determining the weight corresponding to the target third characteristic value interval;

(9) Weighting calculation is carried out on the scores of various second target characteristics of the same client at the downstream of the affinity characteristics of the label layer and the weight corresponding to the target third characteristic value interval to which the score belongs, so that the score of the affinity characteristics of each client is obtained;

the processes of (7) to (9) are similar to the processes of (1) to (3) and will not be described again.

(10) Acquiring fourth characteristic value intervals corresponding to the characteristics of the intimacy of the label layer and weights corresponding to the fourth characteristic value intervals;

(11) Determining a target fourth characteristic value interval to which the score of the characteristic of the intimacy degree of each client belongs, and determining the weight corresponding to the target fourth characteristic value interval;

(12) And performing weighted calculation on the scores of the features of the intimacy degree of each customer and the weights corresponding to a target fourth feature value interval to which the scores belong, and obtaining the intimacy degree of each customer and the current bank according to a weighted calculation result, wherein the first feature value interval corresponding to each feature of the index layer, the weight corresponding to each first feature value interval, the second feature value interval corresponding to each first target feature of the sub-criterion layer, the weight corresponding to each second feature value interval, the third feature value interval corresponding to each second target feature of the criterion layer, the weight corresponding to each third feature value interval, the fourth feature value interval corresponding to the features of the intimacy degree of the label layer and the weights corresponding to each fourth feature value interval are obtained by training according to training data.

The processes of (10) to (12) described above are also similar to the processes of (1) to (3) described above, and will not be described further herein.

Different from the above (1) to (3), after the weighting calculation is performed to obtain the result of the weighting calculation, the affinity between each client and the current bank is represented as a value in 1-5 levels according to the size of the result of the weighting calculation, for example, the result of the weighting calculation belongs to an affinity interval of level 1, that is, the affinity between the client and the current bank is level 1, and the higher the level is, the more the affinity is with the current bank.

In an alternative embodiment of the invention, the method further comprises the steps of:

(1) Acquiring training data, wherein the training data is historical customer service data in a target historical time range, and the target historical time range and the historical time range are separated by a preset time length;

specifically, the duration of the target historical time range is also one month, the preset duration of the interval between the target historical time range and the historical time range may be 3 months, and if the preset duration is 3 months, the predicted conversion probability of each client is the conversion probability of each client in the next two months.

For example, the customer service data in the historical time range is the customer service data of month 6, and then the training data is the customer service data of month 3 (because the customer service data of month 3, looking at the conversion of month 2 in the future, that is, the customer service data of month 4 and the customer service data of month 5, for month 6 currently, the customer service data of month 4 and the customer service data of month 5 can both be obtained, that is, the truth value labels of whether each customer converts in month 4 and month 5 in the customer service data of month 3 can be known), that is, the customer service data of month 3 is used as a training sample to train, and after the training is completed, the customer service data of month 6 is predicted, looking at the conversion of the customer of month 6 in the future of month 2.

(2) Determining initial values of various characteristics of each historical client according to historical client service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment tree diagram;

the determination process is the same as the example in step S102, and is not described herein again. For example, the initial value of APP activity of the E history client is 25, the initial value of APP activity of the f history client is 5, the initial value of APP activity of the g history client is 70, and so on, which are actually a large number of history clients.

(3) Sequencing all history clients according to an initial value of the same characteristic, and clustering the sequenced history clients;

for example, sorting the history clients according to the initial value of the APP liveness to obtain G history clients, E history clients and F history clients, and clustering the sorted history clients.

(4) Determining first characteristic value intervals corresponding to the current characteristics according to the clusters obtained by clustering, and determining the weight corresponding to each first characteristic value interval;

specifically, if 3 clusters are obtained after clustering, wherein the APP activity of all the historical clients in the first cluster is [0, 30], the APP activity of all the historical clients in the second cluster is (30, 60], and the APP activity of all the historical clients in the third cluster is (60, + ∞), it can be determined that the first characteristic value intervals corresponding to the APP activities are [0, 30], (30, 60], (60, + ∞), and further, the weights corresponding to the above [0, 30], (30, 60], (60, + infinity) are determined according to some business rules.

(5) Determining a target first characteristic value interval to which initial values of various characteristics of the same historical client downstream of a first target characteristic of a sub-criterion layer belong, and determining a weight corresponding to the target first characteristic value interval, wherein the first target characteristic is any one of the characteristics of the sub-criterion layer;

the specific process may refer to the intimacy-based assessment tree diagram and the initial values of various characteristics of each client to perform the related description of (2) in the intimacy assessment, which is not described herein again.

(6) Weighting and calculating initial values of various characteristics of the same historical client at the downstream of the first target characteristic of the sub-criterion layer and weights corresponding to target first characteristic value intervals to which the initial values belong to obtain scores of the first target characteristics of the historical clients;

the specific process may refer to the intimacy-based assessment tree diagram and the initial values of various characteristics of each client to perform the related description of (3) in the intimacy assessment, which is not described herein again.

(7) Sequencing the historical clients according to the scores of the same first target characteristic, and clustering the sequenced historical clients;

(8) Determining second characteristic value intervals corresponding to the current first target characteristics according to the clusters obtained by clustering, and determining weights corresponding to the second characteristic value intervals;

(9) Determining a target second characteristic value interval to which the scores of various first target characteristics of the same historical client downstream of a second target characteristic of the criterion layer belong, and determining the weight corresponding to the target second characteristic value interval, wherein the second target characteristic is any one of the characteristics of the criterion layer;

(10) Weighting and calculating the scores of various first target characteristics of the same historical client at the downstream of the second target characteristic of the criterion layer and the weights corresponding to the target second characteristic value intervals to which the scores belong to obtain the scores of the second target characteristics of the historical clients;

(11) Sequencing the historical clients according to the scores of the same second target characteristic, and clustering the sequenced historical clients;

(12) Determining a third characteristic value interval corresponding to the current second target characteristic according to the cluster obtained by clustering, and determining the weight corresponding to each third characteristic value interval;

(13) Determining a target third characteristic value interval to which the scores of various second target characteristics of the same historical client at the downstream of the characteristics of the intimacy of the label layer belong, and determining the weight corresponding to the target third characteristic value interval;

(14) Weighting and calculating the scores of various second target characteristics of the same history client at the downstream of the affinity characteristics of the label layer and the weight corresponding to the target third characteristic value interval to which the score belongs to obtain the score of the affinity characteristics of each history client;

(15) Sequencing each history client according to the score of the feature of the intimacy degree, and clustering each sequenced history client;

(16) And determining fourth characteristic value intervals corresponding to the characteristics of the intimacy degree according to the clusters obtained by clustering, and determining the weight corresponding to each fourth characteristic value interval.

The contents of (7) to (16) above can refer to the corresponding related descriptions, and are not described herein again.

In an optional embodiment of the invention, the method further comprises:

(1) Determining a target fourth characteristic value interval to which the score of the characteristic of the intimacy degree of each historical client belongs, and determining the weight corresponding to the target fourth characteristic value interval;

(2) Weighting calculation is carried out on the scores of the features of the intimacy degree of each history client and the weights corresponding to the target fourth feature value intervals to which the scores belong, and intimacy degree of each history client and the current bank is obtained according to the results of the weighting calculation;

(3) Taking the initial value of each characteristic of each history client, the score of the first target characteristic of each history client, the score of the second target characteristic of each history client and the score of the characteristic of the intimacy of each history client as the intimacy correlation factor of each history client;

(4) Determining basic attributes of each historical client and relevant attributes of a client agent of each historical client according to historical client service data;

(5) And training the initial silver-life customer conversion prediction model through the intimacy between each historical customer and the current bank, the intimacy correlation factor of each historical customer, the basic attribute of each historical customer, the correlation attribute of the customer agent of each historical customer and the label of whether each historical customer converts to obtain the silver-life customer conversion prediction model.

In an optional embodiment of the present invention, the affinity assessment tree is obtained by analyzing and constructing affinity and customer service data in advance.

Specifically, an affinity evaluation dendrogram is constructed from top to bottom by applying an analytic hierarchy process, and a hierarchy (comprising a positive direction and a negative direction) and corresponding weights of all characteristic intervals are set by combining business experience. Further, clustering standardization processing is carried out on the performance of the client in each index (30 + index), and further the intimacy degree score of the client is calculated in an upward summarizing mode according to the weight.

The invention uses the mass business data of the financial institutions to create the initial value of the characteristics of the index layer, breaks through the information barrier based on the intimacy assessment tree diagram, and realizes the description of the relationship between the client and the current bank 'intimacy' from multiple dimensions.

In an alternative embodiment of the invention, the conversion probabilities for each customer include: probability that the life insurance non-bank class user client is converted into the current bank class user client.

Specifically, the conversion probability of each client includes: the probability that a non-bank class of user client of the life insurance will be converted into a current class of user client of the bank in the next two months.

The thermodynamic diagrams of the customers in several dimensions of service penetration, interaction activation, quality experience and negative influence are shown in fig. 3, and it can be seen that the correlation relationships of the service penetration, interaction activation, quality experience and negative influence in terms of intimacy degree do not conflict, and the relationships of intimacy degree and contract number, held product type number, VIP coverage and the like are shown in fig. 4, and it can be seen that the higher intimacy degree, the more contract number, the higher intimacy degree, the more held product type number and the like are shown, and the relationships of intimacy degree and AUM and the like are shown in fig. 5. Through the 3 figures, it can be seen that the intimacy label provided by the invention is accurate and scientific in evaluating intimacy between the customer and the current bank.

In addition, the inventor compares the accuracy of the scheme of the invention with that of the traditional scheme, and finds that the historical prediction accuracy is between 1.5% and 2.5% when the silver-life customer conversion rate is predicted under the traditional expert rule, and the accuracy of 5% can be achieved by using a machine learning model of xgboost/lr. The accuracy rate obtained by utilizing the intimacy factor + lightGBM model framework provided by the invention is 12%, 5 times of the expert rule and 2 times of the xgboost/lr model, and the recall rate is larger than the xgboost/lr model, although the recall rate is not a main consideration index; in addition, the invention has less memory consumption and training time, and is easier to be deployed to the distribution environment of big data

The prediction model based on the intimacy label + + lightGBM is successfully applied to the project of bilateral cooperation of the bank and the life insurance, the predicted life insurance client is issued to the life insurance related agent to be used as a client for getting, touching and triggering business opportunities, the issued amount of the client is 5 ten thousand per month, the annual accumulated client is issued by approximately 60 ten thousand, 7 million new clients of the bank are brought, and the AUM of the bank is increased by 20 hundred million.

Example two:

the embodiment of the invention further provides a device for predicting the silver life customer conversion, which is mainly used for executing the method for predicting the silver life customer conversion provided by the first embodiment of the invention, and the device for predicting the silver life customer conversion provided by the embodiment of the invention is specifically introduced below.

Fig. 6 is a schematic diagram of an apparatus for predicting the conversion of a silver-based customer according to an embodiment of the present invention, as shown in fig. 6, the apparatus mainly includes: an acquisition and determination unit 10, an intimacy degree evaluation unit 20, a determination unit 30, and a silver-life customer conversion prediction unit 40, wherein:

the acquiring and determining unit is used for acquiring client service data in a historical time range and determining initial values of various characteristics of each client according to the client service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment dendrogram, and the intimacy assessment dendrogram comprises the following steps from top to bottom: the system comprises a label layer, a criterion layer, a sub-criterion layer and an index layer;

the intimacy evaluation unit is used for carrying out intimacy evaluation on the basis of the intimacy evaluation tree diagram and initial values of various characteristics of each client to obtain intimacy of each client and a current bank and intimacy related factors of each client, wherein the intimacy related factors are scores of various characteristics contained in a label layer, a criterion layer, a sub-criterion layer and an index layer in the intimacy evaluation tree diagram;

a determining unit, for determining the basic attribute of each client and the related attribute of the client agent of each client according to the client service data;

and the silver-life customer conversion prediction unit is used for inputting the intimacy degree of each customer and the current bank, the intimacy degree correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into the trained silver-life customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver-life customer conversion prediction model is a lightGBM model.

The predicting device for the silver life customer conversion obtains nonlinear intimacy degree of each customer and the current bank and parameters of intimacy degree correlation factors of each customer based on customer service data, and then inputs the intimacy degree of each customer and the current bank, the intimacy degree correlation factors of each customer, basic attributes of each customer and correlation attributes of customer agents of each customer into a trained silver life customer conversion predicting model to predict the silver life customer conversion.

Optionally, the intimacy assessment unit is further configured to: acquiring first characteristic value intervals corresponding to various characteristics of the index layer and weights corresponding to the first characteristic value intervals; determining a target first characteristic value interval to which initial values of various characteristics of the same client downstream of the first target characteristic of the sub-criterion layer belong, and determining a weight corresponding to the target first characteristic value interval, wherein the first target characteristic is any one of the characteristics of the sub-criterion layer; weighting and calculating initial values of various characteristics of the same client at the downstream of the first target characteristic of the sub-criterion layer and weights corresponding to target first characteristic value intervals to which the initial values belong to obtain scores of the first target characteristic of each client; acquiring second characteristic value intervals corresponding to various first target characteristics of the sub-criterion layer and weights corresponding to the second characteristic value intervals; determining a target second characteristic value interval to which the scores of various first target characteristics of the same client downstream of a second target characteristic of the criterion layer belong, and determining the weight corresponding to the target second characteristic value interval, wherein the second target characteristic is any one of the characteristics of the criterion layer; weighting and calculating the scores of various first target characteristics of the same client at the downstream of the second target characteristics of the criterion layer and the weights corresponding to the target second characteristic value intervals to which the scores belong to obtain the scores of the second target characteristics of each client; acquiring third characteristic value intervals corresponding to various second target characteristics of the criterion layer and weights corresponding to the third characteristic value intervals; determining a target third characteristic value interval to which the scores of various second target characteristics of the same customer downstream of the characteristics of the intimacy degree of the label layer belong, and determining the weight corresponding to the target third characteristic value interval; weighting calculation is carried out on the scores of various second target characteristics of the same client at the downstream of the affinity characteristics of the label layer and the weight corresponding to the target third characteristic value interval to which the score belongs, so that the score of the affinity characteristics of each client is obtained; acquiring fourth characteristic value intervals corresponding to the characteristics of the intimacy of the label layer and weights corresponding to the fourth characteristic value intervals; determining a target fourth characteristic value interval to which the score of the characteristic of the intimacy degree of each client belongs, and determining the weight corresponding to the target fourth characteristic value interval; and performing weighted calculation on the scores of the features of the intimacy degree of each customer and the weights corresponding to a target fourth feature value interval to which the scores belong, and obtaining the intimacy degree of each customer and the current bank according to a weighted calculation result, wherein the first feature value interval corresponding to each feature of the index layer, the weight corresponding to each first feature value interval, the second feature value interval corresponding to each first target feature of the sub-criterion layer, the weight corresponding to each second feature value interval, the third feature value interval corresponding to each second target feature of the criterion layer, the weight corresponding to each third feature value interval, the fourth feature value interval corresponding to the features of the intimacy degree of the label layer and the weights corresponding to each fourth feature value interval are obtained by training according to training data.

Optionally, the apparatus is further configured to: acquiring training data, wherein the training data is historical customer service data in a target historical time range, and the target historical time range is separated from the historical time range by a preset time length; determining initial values of various characteristics of each historical client according to historical client service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment tree diagram; sequencing the historical clients according to the initial value of the same characteristic, and clustering the sequenced historical clients; determining first characteristic value intervals corresponding to the current characteristics according to the clusters obtained by clustering, and determining the weight corresponding to each first characteristic value interval; determining a target first characteristic value interval to which initial values of various characteristics of the same historical client downstream of a first target characteristic of a sub-criterion layer belong, and determining a weight corresponding to the target first characteristic value interval, wherein the first target characteristic is any one of the characteristics of the sub-criterion layer; weighting and calculating initial values of various characteristics of the same historical client at the downstream of the first target characteristic of the sub-criterion layer and weights corresponding to target first characteristic value intervals to which the initial values belong to obtain scores of the first target characteristics of the historical clients; sequencing the historical clients according to the scores of the same first target characteristic, and clustering the sequenced historical clients; determining second characteristic value intervals corresponding to the current first target characteristics according to the clusters obtained by clustering, and determining weights corresponding to the second characteristic value intervals; determining a target second characteristic value interval to which the scores of various first target characteristics of the same historical client downstream of a second target characteristic of the criterion layer belong, and determining the weight corresponding to the target second characteristic value interval, wherein the second target characteristic is any one of the characteristics of the criterion layer; weighting and calculating the scores of various first target characteristics of the same historical client at the downstream of the second target characteristic of the criterion layer and the weights corresponding to the target second characteristic value intervals to which the scores belong to obtain the scores of the second target characteristics of the historical clients; sequencing the historical clients according to the scores of the same second target characteristic, and clustering the sequenced historical clients; determining a third characteristic value interval corresponding to the current second target characteristic according to the cluster obtained by clustering, and determining the weight corresponding to each third characteristic value interval; determining a target third characteristic value interval to which the scores of various second target characteristics of the same historical client at the downstream of the characteristics of the intimacy of the label layer belong, and determining the weight corresponding to the target third characteristic value interval; weighting and calculating the scores of various second target characteristics of the same history client at the downstream of the affinity characteristics of the label layer and the weights corresponding to the target third characteristic value intervals to which the scores belong to obtain the scores of the affinity characteristics of the history clients; sequencing the historical clients according to the scores of the characteristics of the intimacy degree, and clustering the sequenced historical clients; and determining fourth characteristic value intervals corresponding to the characteristics of the intimacy degree according to the clusters obtained by clustering, and determining the weight corresponding to each fourth characteristic value interval.

Optionally, the apparatus is further configured to: determining a target fourth characteristic value interval to which the score of the characteristic of the intimacy degree of each historical client belongs, and determining the weight corresponding to the target fourth characteristic value interval; weighting calculation is carried out on the scores of the features of the intimacy degree of each history client and the weights corresponding to the target fourth feature value intervals to which the scores belong, and the intimacy degree of each history client and the current bank is obtained according to the results of the weighting calculation; taking the initial value of each characteristic of each history client, the score of the first target characteristic of each history client, the score of the second target characteristic of each history client and the score of the characteristic of the intimacy of each history client as the intimacy correlation factor of each history client; determining basic attributes of each historical client and relevant attributes of a client agent of each historical client according to historical client service data; and training the initial bank life customer conversion prediction model through the intimacy degree of each history customer and the current bank, the intimacy degree correlation factor of each history customer, the basic attribute of each history customer, the correlation attribute of a customer agent of each history customer and the label of whether each history customer converts or not, so as to obtain the bank life customer conversion prediction model.

Optionally, the affinity assessment dendrogram is obtained by analyzing and constructing affinity and customer service data in advance.

Optionally, the conversion probability of each customer includes: probability that the life insurance non-bank class user client is converted into the current bank class user client.

The device provided by the embodiment of the present invention has the same implementation principle and technical effect as the method embodiments, and for the sake of brief description, reference may be made to the corresponding contents in the method embodiments without reference to the device embodiments.

As shown in fig. 7, an electronic device 600 provided in an embodiment of the present application includes: a processor 601, a memory 602 and a bus, wherein the memory 602 stores machine-readable instructions executable by the processor 601, when the electronic device runs, the processor 601 and the memory 602 communicate with each other through the bus, and the processor 601 executes the machine-readable instructions to execute the steps of the prediction method for the silver-life customer transformation as described above.

Specifically, the memory 602 and the processor 601 can be general-purpose memories and processors, which are not limited in particular, and the prediction method of the silver-life customer conversion can be performed when the processor 601 runs a computer program stored in the memory 602.

The processor 601 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 601. The Processor 601 may be a general-purpose Processor, and includes a Central Processing Unit (CPU), a Network Processor (NP), and the like; the Integrated Circuit may also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, or discrete hardware components. The various methods, steps, and logic blocks disclosed in the embodiments of the present application may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules may be located in ram, flash, rom, prom, or eprom, registers, etc. as is well known in the art. The storage medium is located in the memory 602, and the processor 601 reads the information in the memory 602 and completes the steps of the method in combination with the hardware thereof.

In response to the foregoing method for predicting silver-based customer conversion, an embodiment of the present application further provides a computer-readable storage medium, where the computer-readable storage medium stores machine-executable instructions, and when the computer-executable instructions are called and executed by a processor, the computer-executable instructions cause the processor to execute the foregoing method for predicting silver-based customer conversion.

The prediction device for the silver-life customer conversion provided by the embodiment of the application can be specific hardware on the device or software or firmware installed on the device. The device provided in the embodiment of the present application has the same implementation principle and the same technical effects as those of the foregoing method embodiments, and for the sake of brief description, reference may be made to corresponding contents in the foregoing method embodiments for the absence of any mention in the device embodiment. It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the foregoing systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed coupling or direct coupling or communication connection between each other may be through some communication interfaces, indirect coupling or communication connection between devices or units, and may be in an electrical, mechanical or other form.

For another example, the flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments provided in the present application may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions may be stored in a computer-readable storage medium if they are implemented in the form of software functional units and sold or used as separate products. Based on such understanding, the technical solution of the present application or portions thereof that substantially contribute to the prior art may be embodied in the form of a software product stored in a storage medium and including instructions for causing an electronic device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the vehicle marking method according to the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus once an item is defined in one figure, it need not be further defined and explained in subsequent figures, and moreover, the terms "first", "second", "third", etc. are used merely to distinguish one description from another and are not to be construed as indicating or implying relative importance.

Finally, it should be noted that: the above-mentioned embodiments are only specific embodiments of the present application, and are used to illustrate the technical solutions of the present application, but not to limit the technical solutions, and the scope of the present application is not limited to the above-mentioned embodiments, although the present application is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope disclosed in the present application; such modifications, changes or substitutions do not depart from the scope of the embodiments of the present application. Are intended to be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for predicting a conversion of a silver-life customer, comprising:

and inputting the intimacy degree of each customer and the current bank, the intimacy degree correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into a trained silver customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver customer conversion prediction model is a lightGBM model.

2. The prediction method according to claim 1, wherein performing affinity evaluation based on the affinity evaluation dendrogram and initial values of various features of the clients comprises:

acquiring first characteristic value intervals corresponding to various characteristics of the index layer and weights corresponding to the first characteristic value intervals;

determining a target first characteristic value interval to which initial values of various characteristics of the same client downstream of a first target characteristic of the sub-criterion layer belong, and determining a weight corresponding to the target first characteristic value interval, wherein the first target characteristic is any one of the characteristics of the sub-criterion layer;

weighting and calculating initial values of various characteristics of the same client at the downstream of the first target characteristic of the sub-criterion layer and weights corresponding to a target first characteristic value interval to which the initial values belong to obtain a score of the first target characteristic of each client;

acquiring second characteristic value intervals corresponding to various first target characteristics of the sub-criterion layer and weights corresponding to the second characteristic value intervals;

determining a target second characteristic value interval to which the scores of various first target characteristics of the same customer downstream of a second target characteristic of the criterion layer belong, and determining the weight corresponding to the target second characteristic value interval, wherein the second target characteristic is any one of the characteristics of the criterion layer;

weighting and calculating the scores of various first target characteristics of the same client at the downstream of the second target characteristics of the criterion layer and the weights corresponding to the target second characteristic value intervals to which the scores belong to obtain the scores of the second target characteristics of the clients;

acquiring third characteristic value intervals corresponding to various second target characteristics of the criterion layer and weights corresponding to the third characteristic value intervals;

determining a target third characteristic value interval to which scores of various second target characteristics of the same customer downstream of the characteristics of the intimacy degree of the label layer belong, and determining the weight corresponding to the target third characteristic value interval;

weighting and calculating the scores of various second target characteristics of the same client at the downstream of the affinity characteristics of the label layer and the weights corresponding to the target third characteristic value intervals to which the scores belong to obtain the scores of the affinity characteristics of each client;

acquiring fourth characteristic value intervals corresponding to the characteristics of the intimacy degree of the label layer and weights corresponding to the fourth characteristic value intervals;

determining a target fourth characteristic value interval to which the score of the characteristic of the intimacy degree of each customer belongs, and determining the weight corresponding to the target fourth characteristic value interval;

and performing weighted calculation on the score of the feature of the intimacy degree of each customer and the weight corresponding to the target fourth feature value interval to which the score belongs, and obtaining the intimacy degree of each customer and the current bank according to the result of the weighted calculation, wherein the first feature value interval corresponding to each feature of the index layer, the weight corresponding to each first feature value interval, the second feature value interval corresponding to each first target feature of the sub-criterion layer, the weight corresponding to each second feature value interval, the third feature value interval corresponding to each second target feature of the criterion layer, the weight corresponding to each third feature value interval, the fourth feature value interval corresponding to the feature of the intimacy degree of the label layer, and the weight corresponding to each fourth feature value interval are obtained by training according to training data.

3. The prediction method according to claim 1, characterized in that the method further comprises:

acquiring training data, wherein the training data is historical customer service data in a target historical time range, and the target historical time range and the historical time range are separated by a preset time length;

determining initial values of various characteristics of each historical client according to the historical client service data, wherein the various characteristics are characteristics of an index layer in an intimacy assessment tree diagram;

sequencing all history clients according to an initial value of the same characteristic, and clustering the sequenced history clients;

determining a first characteristic value interval corresponding to the current characteristic according to the cluster obtained by clustering, and determining the weight corresponding to each first characteristic value interval;

determining a target first characteristic value interval to which initial values of various characteristics of the same historical client downstream of a first target characteristic of the sub-criterion layer belong, and determining a weight corresponding to the target first characteristic value interval, wherein the first target characteristic is any one of the characteristics of the sub-criterion layer;

weighting and calculating initial values of various characteristics of the same historical client at the downstream of the first target characteristic of the sub-criterion layer and weights corresponding to target first characteristic value intervals to which the initial values belong to obtain scores of the first target characteristics of the historical clients;

sequencing the historical clients according to the scores of the same first target characteristic, and clustering the sequenced historical clients;

determining second characteristic value intervals corresponding to the current first target characteristics according to the clusters obtained by clustering, and determining the weight corresponding to each second characteristic value interval;

determining a target second characteristic value interval to which scores of various first target characteristics of the same historical client downstream of a second target characteristic of the criterion layer belong, and determining a weight corresponding to the target second characteristic value interval, wherein the second target characteristic is any one of the characteristics of the criterion layer;

weighting and calculating the scores of various first target characteristics of the same historical client at the downstream of the second target characteristic of the criterion layer and the weights corresponding to the target second characteristic value intervals to which the scores belong to obtain the scores of the second target characteristics of the historical clients;

sequencing the historical clients according to the scores of the same second target characteristic, and clustering the sequenced historical clients;

determining a third characteristic value interval corresponding to the current second target characteristic according to the cluster obtained by clustering, and determining the weight corresponding to each third characteristic value interval;

determining a target third characteristic value interval to which scores of various second target characteristics of the same history customer downstream of the characteristics of the intimacy of the label layer belong, and determining the weight corresponding to the target third characteristic value interval;

weighting and calculating the scores of various second target characteristics of the same historical client at the downstream of the affinity characteristics of the label layer and the weights corresponding to the target third characteristic value intervals to which the scores belong to obtain the scores of the affinity characteristics of the historical clients;

sequencing each history client according to the score of the feature of the intimacy degree, and clustering each sequenced history client;

and determining fourth characteristic value intervals corresponding to the characteristics of the intimacy degree according to the clusters obtained by clustering, and determining the weight corresponding to each fourth characteristic value interval.

4. The prediction method according to claim 3, characterized in that the method further comprises:

determining a target fourth characteristic value interval to which the score of the characteristic of the intimacy degree of each historical client belongs, and determining the weight corresponding to the target fourth characteristic value interval;

weighting calculation is carried out on the scores of the features of the intimacy degree of each history client and the weights corresponding to the target fourth feature value intervals to which the scores belong, and the intimacy degree of each history client and the current bank is obtained according to the results of the weighting calculation;

setting the initial value of each characteristic of each history client, the score of the first target characteristic of each history client, the score of the second target characteristic of each history client and the score of the characteristic of the intimacy of each history client as the intimacy degree correlation factor of each history client;

determining basic attributes of each historical client and relevant attributes of client agents of each historical client according to the historical client service data;

and training an initial silver life customer conversion prediction model through the intimacy degree of each history customer and the current bank, the intimacy degree correlation factor of each history customer, the basic attribute of each history customer, the correlation attribute of a customer agent of each history customer and the label of whether each history customer converts or not, so as to obtain the silver life customer conversion prediction model.

5. The prediction method according to claim 1, wherein the intimacy assessment dendrogram is obtained by analyzing and constructing intimacy and customer service data in advance.

6. The prediction method according to claim 1, wherein the conversion probability for each customer comprises: probability that the life insurance non-bank class user client is converted into the current bank class user client.

7. A prediction apparatus for silver-life customer conversion, comprising:

an affinity evaluation unit, configured to perform affinity evaluation based on the affinity evaluation dendrogram and initial values of various features of each client to obtain affinities of each client and a current bank and affinity correlation factors of each client, where the affinity correlation factors are scores of the features included in the tag layer, the criterion layer, the sub-criterion layer, and the index layer in the affinity evaluation dendrogram;

and the silver life customer conversion prediction unit is used for inputting the intimacy degree of each customer and the current bank, the intimacy degree correlation factor of each customer, the basic attribute of each customer and the correlation attribute of the customer agent of each customer into the trained silver life customer conversion prediction model to obtain the conversion probability of each customer, wherein the silver life customer conversion prediction model is a lightGBM model.

8. The prediction apparatus according to claim 7, wherein the intimacy degree evaluation unit is further configured to:

weighting and calculating initial values of various characteristics of the same client at the downstream of the first target characteristic of the sub-criterion layer and weights corresponding to target first characteristic value intervals to which the initial values belong to obtain scores of the first target characteristics of the clients;

determining a target second characteristic value interval to which the scores of various first target characteristics of the same customer downstream of a second target characteristic of the criterion layer belong, and determining the corresponding weight of the target second characteristic value interval, wherein the second target characteristic is any one of the characteristics of the criterion layer;

weighting and calculating the scores of various first target characteristics of the same client at the downstream of the second target characteristic of the criterion layer and the weights corresponding to the target second characteristic value intervals to which the scores belong to obtain the scores of the second target characteristics of the clients;

acquiring a fourth characteristic value interval corresponding to the characteristics of the intimacy of the label layer and the weight corresponding to each fourth characteristic value interval;

determining a target fourth characteristic value interval to which the score of the characteristic of the intimacy degree of each client belongs, and determining the weight corresponding to the target fourth characteristic value interval;

9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the method of any of the preceding claims 1 to 6 are implemented when the computer program is executed by the processor.

10. A computer readable storage medium, having stored thereon machine executable instructions, which when invoked and executed by a processor, cause the processor to perform the method of any of claims 1 to 6.