CN116522102A - Method and device for dynamically adjusting accident risk model database based on privacy calculation - Google Patents

Abstract

The application discloses a method and a device for dynamically adjusting an accident risk model database based on privacy calculation, which relate to the technical field of databases and are used for acquiring accident risk modeling samples of insurance company clients; carrying out privacy intersection on the accident risk modeling sample to obtain an intersection sample; according to the label library of the intersection sample associated data source side, obtaining independent variable information and dependent variable information; calculating the average IV value of the tag library according to the independent variable information and the dependent variable information; and dynamically adjusting the ordering and pricing mechanism of the tag library according to the average IV value. According to the method and the device for dynamically adjusting the accident risk model database based on privacy calculation, under the condition that supervision is guaranteed, the tag library is dynamically optimized through the average IV value, so that the value of a data source is fully mined, safety and reliability are achieved, and the cost is low and the gain is high.

Description

Method and device for dynamically adjusting accident risk model database based on privacy calculation

technical field

The present application relates to the field of database technology, in particular to a method and device for dynamically adjusting an accident risk model database based on privacy calculation.

Background technique

Accidental injury insurance (abbreviation: accident insurance) is one of the personal insurance businesses. Personal insurance that pays insurance money based on the death or disability of the insured due to accidental injury. Its basic content is: the insured pays a certain premium to the insurer, if the insured suffers accidental injury within the insurance period and takes this as the direct cause or proximate cause, within a certain period of time from the date of the accidental injury caused The insurer will pay the insured or its beneficiaries a certain amount of insurance money if the insured dies, becomes disabled, pays medical expenses or temporarily loses the ability to work. There are two protection items, namely death benefit and disability benefit.

With the continuous development of society, the trend of insurance companies to construct pre-insurance risk control models based on external data sources has formed. When a traditional insurance company seeks external data to enhance its risk control model, it will provide a part of three-factor encrypted labeled samples to the data source, and let the modeling engineer of the data source provide modeling services, so as to finally obtain an expected The model is finally deployed in the data source to provide services to insurance companies in the form of APIs.

However, with the promulgation of the Data Security Law and related regulations on the protection of personal privacy information, direct access to personal encrypted information does not meet the requirements of future supervision. Considering the product structure, if only by maintaining fixed data assets and providing sample modeling The service cannot fully reflect the value of data originating from existing data, and the data source lacks insight into the target customer's financial history, shopping preferences, identity traits, and personal connections.

Contents of the invention

For this reason, this application provides a method and device for dynamically adjusting the accident risk model database based on privacy calculations, so as to solve the existing problems in the prior art that the data source cannot fully reflect the value of data originating from existing data, and lacks multi-dimensionality. The problem of poor insight and poor security.

In order to achieve the above object, the application provides the following technical solutions:

In the first aspect, a method for dynamically adjusting the accident risk model database based on privacy calculation, including:

Obtain modeling samples of insurance company customers' accident insurance;

Perform privacy intersecting on the accident risk modeling samples to obtain the intersection samples;

Obtain independent variable information and dependent variable information according to the tag library on the data source side associated with the intersection sample;

Calculate the average IV value of the tag library according to the independent variable information and dependent variable information;

Dynamically adjust the sorting and pricing mechanism of the tag library according to the average IV value.

Preferably, the accident risk modeling sample includes at least name, ID number, mobile phone number and backtracking time.

Preferably, a left-association query is used when associating the tag library at the data source side according to the intersection samples.

Preferably, the label library includes a warranty library, a shopping preference library, a reverse selection group library and an overdue library.

Preferably, when dynamically adjusting the sorting and pricing mechanism of the tag library according to the average IV value, the greater the average IV value, the greater the corresponding tag library weight or exposure value.

In the second aspect, a method for dynamically adjusting the accident risk model database based on privacy calculations, including:

Obtain modeling samples of insurance company customers' accident insurance;

Perform privacy intersecting on the accident risk modeling samples to obtain the intersection samples;

Obtain independent variable information and dependent variable information according to the tag library on the data source side associated with the intersection sample;

Calculate the missing rate of the tag library according to the independent variable information and dependent variable information;

Dynamically adjust the sorting and pricing mechanism of the tag library according to the missing rate.

Preferably, when dynamically adjusting the sorting and pricing mechanism of the tag library according to the missing rate, the higher the missing rate, the smaller the corresponding tag library weight or exposure value.

In the third aspect, a device for dynamically adjusting the accident risk model database based on privacy calculations, including:

Data upload module, which is used to obtain modeling samples of insurance company customers' accident insurance;

A privacy intersecting module, configured to perform privacy intersecting on the accident risk modeling samples to obtain intersection samples;

The tag library processing module is used to obtain independent variable information and dependent variable information according to the tag library on the data source side associated with the intersection sample;

Privacy statistics module, for calculating the average IV value of the tag library according to the independent variable information and dependent variable information;

And dynamically adjust the sorting and pricing mechanism of the tag library according to the average IV value.

In a fourth aspect, a computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the method for dynamically adjusting the accident risk model database based on privacy calculation when executing the computer program.

In the fifth aspect, a computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, the steps of the method for dynamically adjusting the accident risk model database based on privacy calculation are realized.

Compared with the prior art, the present application has at least the following beneficial effects:

This application provides a method and device for dynamically adjusting the accident insurance model database based on privacy calculations, by obtaining accident insurance modeling samples from insurance company customers; performing privacy calculation on the accident insurance modeling samples to obtain intersection samples; The tag library on the data source side obtains independent variable information and dependent variable information; calculates the average IV value of the tag library according to the independent variable information and dependent variable information; dynamically adjusts the sorting and pricing mechanism of the tag library according to the average IV value. The method and device for dynamically adjusting the accident risk model database based on privacy calculations provided by this application can dynamically optimize the label library through the average IV value under the condition of ensuring compliance with supervision, and fully tap the value of the data source, which is not only safe and reliable, but also cost-effective Low, high gain.

Description of drawings

In order to illustrate the prior art and the present application more intuitively, several exemplary drawings are given below. It should be understood that the specific shapes and structures shown in the accompanying drawings should generally not be regarded as limiting conditions for the implementation of the present application; for example, those skilled in the art are able to The increase/decrease/attribution division, specific shape, positional relationship, connection mode, size ratio relationship, etc. of some units (parts) are easy to make conventional adjustments or further optimization.

FIG. 1 is a flow chart of a method for dynamically adjusting the accident risk model database based on privacy calculation provided in Embodiment 1 of the present application;

FIG. 2 is a structural block diagram of a device for dynamically adjusting an accident risk model database based on privacy calculation provided in Embodiment 2 of the present application.

Detailed ways

The present application will be further described in detail through specific embodiments below in conjunction with the accompanying drawings.

In the description of this application: unless otherwise specified, "plurality" means two or more. The terms "first", "second", and "third" in this application are intended to distinguish the referred objects, and have no special meaning in terms of technical connotation (for example, it should not be understood as a reference to the degree of importance or order, etc. emphasis). Expressions such as "comprising", "including", and "having" also mean "not limited to" (certain elements, components, materials, steps, etc.).

Terms such as "upper", "lower", "left", "right", "middle", etc. quoted in this application are usually for the convenience of intuitive understanding with reference to the drawings, rather than absolute terms for the positional relationship in the actual product limited. Without departing from the technical concept disclosed in this application, changes in these relative positional relationships should also be regarded as the scope of this application.

Embodiment one

Referring to Fig. 1, this embodiment provides a method for dynamically adjusting the accident risk model database based on privacy calculation, including:

S1: Obtain modeling samples of insurance company customers' accident insurance;

Specifically, the insurance company provides a batch of accident insurance modeling samples M. The accident insurance modeling samples M must contain encrypted name, ID number, and mobile phone number, and include the insurance application time as the starting point of the backtracking variable.

S2: Intersect the accident risk modeling samples in privacy to obtain the intersection samples;

Specifically, privacy computing refers to a collection of technologies that realize data analysis and computing on the premise of protecting the data itself from being leaked to the outside world, so as to achieve the purpose of "available and invisible" to the data. In this step, the premise of conducting privacy calculation on the accident insurance modeling sample M given by the insurance company is that both parties have deployed privacy computing nodes locally, and the intersection sample M’ is obtained after the privacy calculation.

S3: Obtain independent variable information and dependent variable information according to the tag library on the data source side associated with the intersection sample;

Specifically, the label library is the initial library for the accident risk modeling samples provided by customers. The variables in each library table can be assumed to have a certain correlation with the accident risk modeling samples, especially for the accident risk scenario. In the model sample, the insurance agent attributes, overdue attributes, and whether they belong to the reverse selection group of customers in the model sample may be related to the dependent variable (y value) of the sample.

In this embodiment, for the obtained accident risk modeling samples, the data source side initially configures a tag library, which includes A1 insurance agent library, A2 shopping preference library, A3 reverse selection group library...Ai...An, etc. Each tag library contains the corresponding variable list (list). In this step, prepare for the subsequent calculation of corresponding statistical values through the y value of the left-associated data source side, such as indicators such as claims and accidents, and independent variables (x values).

S4: Calculate the average IV value of the tag library according to the independent variable information and the dependent variable information;

Specifically, through the information of the bank variable (independent variable x) and the information of accident and compensation (dependent variable y) associated with step S2, this step can accurately calculate the average IV value of each bank.

IV (information value) is the value of information, which is an index to measure the predictive ability of features for the model, and is often used as a reference for feature screening before model training.

For example: for the A1 insurance agent library, assume that there are 5 variables in the insurance agent library (the number of insurance policies sold by the user as an insurance agent in the past 3/6/9/12/24 months) with an average IV value of 0.2; for the A4 overdue library, assume There are 5 variables in overdue inventory (users have overdue orders in the past 3/6/9/12/24 months), and the average IV value is 0.3; there are 10 variables for An## inventory, and the average IV value is 0.01; as a preference, the weight or exposure value of the A4 overdue library can be maximized in the future, that is, it is most likely to be selected by the customer to enter the model, and the weight can be lowered for the An## library, which can be effective Save storage resources.

S5: Dynamically adjust the sorting and pricing mechanism of the tag library according to the average IV value.

Through the above steps, after different insurance companies provide multiple batches of samples, continuous corrections can be made, so that the database sorting can be continuously optimized based on the IV value as the sorting index. When modeling, the customer can choose the database variable with the highest IV value as the preferred layer, and some variables with poor performance will be dynamically adjusted by the variable with better performance (be behind or directly eliminated), which saves resources on the whole, and at the same time The number of preferred layer variables used can also be used as a reference for subsequent business negotiation, that is, the more preferred layer variables a customer enters into the model, the higher the charge may be.

The method of dynamically adjusting the accident insurance model database based on privacy calculation provided by this embodiment makes the data more secure and compliant through privacy calculation, and the sample provided by the insurance company customer can just become the focus of optimizing the iterative data system, realizing the data assets The efficient and dynamic operation fully exploits the value of data, balances benefits and costs, and turns disadvantages into opportunities for iterative upgrades, with high security, low cost, and high gains.

Embodiment two

This embodiment provides another method for dynamically adjusting the accident risk model database based on privacy calculation, including:

S1: Obtain modeling samples of insurance company customers' accident insurance;

S2: Intersect the accident risk modeling samples in privacy to obtain the intersection samples;

S3: Obtain independent variable information and dependent variable information according to the tag library on the data source side associated with the intersection sample;

S4: Calculate the missing rate of the tag library according to the independent variable information and the dependent variable information;

S5: Dynamically adjust the sorting and pricing mechanism of the tag library according to the missing rate.

This embodiment uses the missing rate to dynamically adjust the sorting and pricing mechanism of the tag library. When dynamically adjusting the sorting and pricing mechanism of the tag library according to the missing rate, the higher the missing rate, the smaller the corresponding tag library weight or exposure value, that is, the missing rate The higher the value, the lower the probability that the customer will be selected when modeling.

Embodiment three

Please refer to Fig. 2, a device for dynamically adjusting the accident risk model database based on privacy calculation in this embodiment, including:

Data upload module, which is used to obtain modeling samples of insurance company customers' accident insurance;

The privacy intersection module is used to perform privacy intersection on accident risk modeling samples to obtain intersection samples;

The tag library processing module is used to obtain independent variable information and dependent variable information according to the tag library on the data source side associated with the intersection sample;

The privacy statistics module is used to calculate the average IV value of the tag library according to the independent variable information and the dependent variable information;

And dynamically adjust the sorting and pricing mechanism of the tag library according to the average IV value.

For specific limitations on a device for dynamically adjusting an accident risk model database based on privacy calculations, please refer to the above definition for a method for dynamically adjusting an accident risk model database based on privacy calculations, which will not be repeated here.

Embodiment four

This embodiment provides a computer device, including a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, the steps of the method for dynamically adjusting the accident risk model database based on privacy calculation are realized.

Embodiment five

This embodiment provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the step of dynamically adjusting the accident risk model database based on privacy calculation is implemented.

The technical features of the above embodiments can be combined arbitrarily (as long as there is no contradiction in the combination of these technical features), for the sake of concise description, all possible combinations of the various technical features in the above embodiments are not described; these are not clear All the written examples should also be regarded as within the scope of the description in this specification.

The present application has been described more specifically and in detail through general descriptions and specific examples above. It should be understood that based on the technical concept of the present application, some conventional adjustments or further innovations can also be made to these specific embodiments; Also fall within the protection scope of the claims of the present application.

Claims (10)

1. A method for dynamically adjusting an accident risk model database based on privacy calculations, comprising:

acquiring an accident risk modeling sample of an insurance company client;

carrying out privacy intersection on the accident risk modeling sample to obtain an intersection sample;

obtaining independent variable information and dependent variable information according to the label library of the intersection sample associated data source side;

calculating the average IV value of the tag library according to the independent variable information and the dependent variable information;

and dynamically adjusting the ordering and pricing mechanism of the tag library according to the average IV value.

2. The method for dynamically adjusting an accident risk model database based on privacy calculations of claim 1, wherein the accident risk modeling samples include at least name, identification number, phone number, and backtracking time.

3. The method for dynamically adjusting an accident risk model database based on privacy calculations of claim 1, wherein a left-associated query is employed when associating a tag library on a data source side according to the intersection sample.

4. The method for dynamically adjusting an accident risk model database based on privacy calculations of claim 1, wherein the tag library comprises a warranty library, a shopping preference library, an inverse selection crowd library, and an overdue library.

5. The method for dynamically adjusting a risk of accident model database based on privacy calculations according to claim 1, wherein when the ordering and pricing mechanism of the tag library is dynamically adjusted according to the average IV value, the larger the average IV value is, the larger the tag library weight or exposure value corresponding to the average IV value is.

6. A method for dynamically adjusting an accident risk model database based on privacy calculations, comprising:

acquiring an accident risk modeling sample of an insurance company client;

carrying out privacy intersection on the accident risk modeling sample to obtain an intersection sample;

obtaining independent variable information and dependent variable information according to the label library of the intersection sample associated data source side;

calculating the deletion rate of the tag library according to the independent variable information and the dependent variable information;

and dynamically adjusting the ordering and pricing mechanism of the tag library according to the deletion rate.

7. The method for dynamically adjusting a risk of accident model database based on privacy calculations of claim 6, wherein when the ordering and pricing mechanism of the tag library is dynamically adjusted according to the deletion rate, the higher the deletion rate is, the smaller the tag library weight or exposure value is.

8. An apparatus for dynamically adjusting an accident risk model database based on privacy calculations, comprising:

the data uploading module is used for acquiring an accident risk modeling sample of an insurance company client;

the privacy intersection module is used for performing privacy intersection on the accident risk modeling sample to obtain an intersection sample;

the tag library processing module is used for obtaining independent variable information and dependent variable information according to the tag library of the intersection sample associated data source side;

the privacy statistics module is used for calculating the average IV value of the tag library according to the independent variable information and the dependent variable information;

and dynamically adjusting the ordering and pricing mechanism of the tag library according to the average IV value.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 5 when the computer program is executed.

10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 5.