Negative sample screening method and device, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a negative sample screening method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring an original insurance data set, and determining a first proportion of the number of positive samples to the number of negative samples in the original insurance data set; determining each sample set corresponding to the original insurance data set, and determining a target negative sample screening rule matched with the target sample set; screening negative samples in the target sample set based on a target negative sample screening rule to obtain a screened target sample set; determining a second proportion of the number of the positive samples to the number of the negative samples in the screened target sample set, and stopping screening the target sample set under the condition that the second proportion meets a preset proportion requirement; the first proportion is smaller than the second proportion. According to the scheme, the number of negative samples in the training data can be reduced, the proportion of the positive samples to the negative samples is increased, and a basis is provided for training to obtain a high-quality recommendation model for potential insurance users.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a negative sample screening method, device, electronic equipment and storage medium. Background Art

[0002] With the advancement of technology and the intensification of market competition, personalized services and precise recommendations have become important trends in the development of the insurance industry today. Recommending potential insurance users through trained machine learning models has been widely used in the industry. However, due to the particularity of the insurance industry, the ratio of positive and negative samples is extremely unbalanced, which can reach 1:500 or even higher. This causes the trained model to be more inclined to capture the characteristics of negative samples and ignore the characteristics of positive samples, making it difficult to recommend potential insurance users through the model.

[0003] How to reduce the number of negative samples in training data and increase the ratio of positive and negative samples, thereby providing a basis for training recommendation models for high-quality potential insurance users, is a key research issue in the industry. Summary of the invention

[0004] The present invention provides a negative sample screening method, device, electronic device and storage medium, which can reduce the number of negative samples in training data, increase the ratio of positive and negative samples, and provide a basis for training a recommendation model for high-quality potential insurance users.

[0005] According to one aspect of the present invention, a method for screening negative samples is provided, the method comprising:

[0006] Acquire an original insurance data set, and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set; wherein the positive samples are insurance data of insured target insurance products, and the negative samples are insurance data of uninsured target insurance products;

[0007] In the case where the first ratio does not meet the preset ratio requirement, determining each sample set corresponding to the original insurance data set, and determining a target negative sample screening rule matching the target sample set;

[0008] Screening each negative sample in the target sample set based on the target negative sample screening rule to obtain a screened target sample set;

[0009] Determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets a preset ratio requirement; the first ratio is less than the second ratio.

[0010] According to another aspect of the present invention, a negative sample screening device is provided, the device comprising:

[0011] A first ratio determination module is used to obtain an original insurance data set and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set; wherein the positive samples are insurance data of insured target insurance products, and the negative samples are insurance data of uninsured target insurance products;

[0012] A sample set determination module, used for determining each sample set corresponding to the original insurance data set and determining a target negative sample screening rule matching the target sample set when the first ratio does not meet the preset ratio requirement;

[0013] A negative sample screening module, which screens each negative sample in the target sample set based on the target negative sample screening rule to obtain a screened target sample set;

[0014] A second ratio determination module is used to determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets a preset ratio requirement; the first ratio is smaller than the second ratio.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] at least one processor; and

[0017] a memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the negative sample screening method described in any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the negative sample screening method described in any embodiment of the present invention when executed.

[0020] According to another aspect of the present invention, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the negative sample screening method according to any embodiment of the present invention is implemented.

[0021] The technical solution of the embodiment of the present invention is to obtain an original insurance data set and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set; wherein the positive samples are insurance data of target insurance products that have been insured, and the negative samples are insurance data of target insurance products that have not been insured; when the first ratio does not meet the preset ratio requirement, determine each sample set corresponding to the original insurance data set, and determine the target negative sample screening rule that matches the target sample set; based on the target negative sample screening rule, screen each negative sample in the target sample set to obtain a screened target sample set; determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets the preset ratio requirement; the first ratio is less than the second ratio, which can reduce the number of negative samples in the training data and increase the ratio of positive and negative samples, and can provide a basis for training a recommendation model for high-quality potential insurance users.

[0022] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 is a flow chart of a negative sample screening method provided according to Embodiment 1 of the present invention;

[0025] Figure 2 is a flow chart of a negative sample screening method provided according to Embodiment 2 of the present invention;

[0026] Figure 3 is a flow chart of another negative sample screening method provided according to the second embodiment of the present invention;

[0027] Figure 4 is a schematic structural diagram of a negative sample screening device provided according to Embodiment 3 of the present invention;

[0028] Figure 5 It is a schematic diagram of the structure of an electronic device for implementing the negative sample screening method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0031] Embodiment 1

[0032] Figure 1 This is a flowchart of a negative sample screening method provided according to the first embodiment of the present invention. This embodiment is applicable to the case of screening negative samples in the original data set. The method can be performed by a negative sample screening device, which can be implemented in the form of hardware and / or software. The negative sample screening device can be configured in electronic devices such as computers, servers or tablet computers. Figure 1 As shown, the method includes:

[0033] Step 110: Obtain an original insurance data set, and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set.

[0034] Among them, the positive sample is the insurance data of the insured target insurance product, and the negative sample is the insurance data of the uninsured target insurance product. Among them, the insurance data may include user information, policy information, claim information or historical transaction records, etc., which are not limited in this embodiment. The target insurance product can be any insurance product, such as major disease insurance, accident insurance, real estate insurance or vehicle insurance, etc., which are not limited in this embodiment. In this embodiment, the original insurance data may include at least one of the following: user information, policy information, claim information and historical transaction records, etc.; wherein, user information may include the user's name, age, mobile phone number or mailing address, etc.; policy information may include specific insurance content or claim settlement plan, etc., which are not limited in this embodiment.

[0035] In this embodiment, the original insurance data set may be historical insurance data collected from various insurance business systems, for example, relevant information of a policyholder of a health insurance; or relevant information of an uninsured user of an accident insurance product, etc., which is not limited in this embodiment.

[0036] In an optional implementation of the present embodiment, obtaining an original insurance data set and determining a first ratio of the number of positive samples to the number of negative samples in the original insurance data set may include: obtaining original insurance data from each target system respectively; preprocessing each of the original insurance data to obtain an original insurance data set; determining positive samples and negative samples in the original insurance data set respectively, and counting the number of positive samples and the number of negative samples; and determining the ratio of the number of positive samples to the number of negative samples as the first ratio.

[0037] The target system is each insurance business system, for example, the core system of an insurance company, a customer relationship management system, a claims system, a customer service center system, etc., which is not limited in this embodiment.

[0038] Optionally, in this embodiment, after the original insurance data is reacquired from each target system, each insurance data may be further preprocessed to obtain an original insurance data set.

[0039] In an optional implementation of this embodiment, preprocessing of each insurance data may include: removing duplicate data, processing missing values, processing abnormal values, processing divergent features, etc., which are not limited in this embodiment.

[0040] In this embodiment, for duplicate data: each original insurance data in the original insurance data set can be traversed to determine the duplicate insurance data, and the duplicate insurance data can be deleted; for missing values, the following strategies can be adopted for processing: statistics such as mean, median, mode, etc. can be used to fill the missing values ​​of numerical data; for categorical data, the most frequent category can be used for filling; the model can also be used to predict missing values; the rows or columns where the missing values ​​are located can also be deleted; for some special cases, such as cases where missing values ​​may contain useful information, the missing values ​​can be processed as a separate category, or a specific filling method can be used, which is not limited in this embodiment.

[0041] For outliers, the following strategies can be adopted: the quality of data can be improved by identifying and deleting outliers. For example, a box plot can be used to detect outliers and delete them; the winsorization method can be used, or outliers can be replaced with reasonable values ​​(such as median, mean, etc.); outliers can also be converted into other forms through data transformation (for example, Log transformation, Box-Cox transformation, Rank transformation, etc.); for divergent data (for example, the value of a certain data changes too much, that is, the divergence is serious, or the divergence is too small, and the empirical variance is less than 0.1): these data are usually discarded.

[0042] For data with strong correlation: In the data set, there may be data with strong correlation, and the Pearson correlation coefficient is between -0.95 and -0.75 or between 0.75 and 0.95, which may affect the stability and interpretability of the model. Therefore, it is necessary to process the data with strong correlation, usually by selecting one of the data to retain, or using principal component analysis and other methods to reduce the dimension; in this embodiment, for data with wrong format or wrong type, corresponding adjustments and processing (for example, modifying the corresponding errors) are required to ensure the consistency and accuracy of the data.

[0043] It should be noted that, in this embodiment, all operations and processing methods in the data cleaning process are documented and reported to facilitate subsequent reproduction and review, and also facilitate communication and exchanges between teams.

[0044] Further, the positive samples and negative samples in the original insurance data set can be determined. For example, the positive samples and negative samples can be determined according to the insurance status of each original insurance data; for example, the insured original insurance data is determined as the positive sample, and the uninsured insurance data is determined as the negative sample; further, the number of positive samples and the number of negative samples can be counted separately; the ratio of the number of positive samples to the number of negative samples is determined, and the ratio is determined as a first ratio; for example, the first ratio can be 1:500, 1:600 ​​or 1:1000, etc., which is not limited in this embodiment.

[0045] It should be noted that, in this embodiment, the user information is obtained only after the user's authorization, and the method of obtaining is reasonable and legal.

[0046] Step 120: When the first ratio does not meet the preset ratio requirement, determine each sample set corresponding to the original insurance data set, and determine a target negative sample screening rule that matches the target sample set.

[0047] In an optional implementation of the present embodiment, after determining a first ratio of the number of positive samples to the number of negative samples in the original insurance data set, it can be further determined whether the first ratio meets a preset ratio requirement, wherein the preset ratio requirement can be a threshold requirement, for example, whether the first ratio is greater than a preset ratio threshold, and the preset ratio threshold can be any value, for example: 1:10 or 1:5, etc., which is not limited in the present embodiment.

[0048] Optionally, in this embodiment, if the first ratio is less than a preset ratio threshold, it can be determined that the first ratio does not meet the preset ratio requirement; further, each sample set corresponding to the original insurance data set can be determined, and the negative sample screening rules matching each sample set can be determined respectively; wherein, the target sample set can be any sample set, which is not limited in this embodiment.

[0049] It can be understood that the original insurance data set is insurance data obtained from different systems, and the insurance products or claims information they target are different. Therefore, in this embodiment, the original insurance data set is divided into different sample sets (exemplarily, each sample set targets an insurance product, or the user information of each sample set is similar, which is not limited in this embodiment). This can facilitate the screening of negative samples and provide a basis for improving the screening efficiency of negative samples.

[0050] Step 130: Screen each negative sample in the target sample set based on the target negative sample screening rule to obtain a screened target sample set.

[0051] In an optional implementation of this embodiment, after determining the target negative sample screening rule that matches the target sample set, the negative samples in the target sample set can be further screened according to the target negative sample screening rule to obtain a screened target sample set.

[0052] Among them, the target screening rules may be product information screening rules, activity information screening rules, and telephone appointment and face-to-face screening rules, etc., which are not limited in this embodiment.

[0053] Optionally, in this embodiment, the target negative sample screening rule can be determined by the following steps: if the target negative sample screening rule is a product information screening rule, then the target insurance customer in the target insurance product is determined, the target insurance customer at any time is determined as the target negative sample, and the target negative sample is deleted; wherein, the target insurance customer is a customer who fails to pay the premium on time and / or is rejected for insurance; if the target negative sample screening rule is an activity information screening rule, then the data characteristics of the target activity are determined, and based on the data characteristics, a first negative sample with a correlation with the target activity lower than a set threshold is determined, the first negative sample is determined as the target negative sample, and the target negative sample is deleted; the data characteristics include attendance data and activity data; if the target negative sample screening rule is a telephone appointment and face-to-face screening rule, then the target user whose telephone appointment failed, or whose telephone appointment succeeded but the contract was not successfully signed is determined, the negative sample corresponding to the target user is determined as the target negative sample, and the target negative sample is deleted.

[0054] In an optional implementation of the present embodiment, if the target negative sample screening rule is a product information screening rule, the target insurance customer in the target insurance product can be further determined, wherein the target insurance customer may be a customer who fails to pay the premium on time or is rejected for insurance, that is, the target insurance customer may not be insured by the target insurance product due to failure to meet the insurance requirements or failure to pay the premium; further, the target insurance customer can be determined as a target negative sample, and further, the target negative sample can be deleted.

[0055] Optionally, in an example of this embodiment, for critical illness insurance products, the target insurance customers (i.e., target negative samples) may select customers who have not exceeded the deductible or have not met the pre-underwriting conditions. These customers may be denied insurance or require additional deductibles because their health conditions do not meet the requirements or the underwriting results do not meet expectations; for increasing term life products, the target insurance customers (i.e., target negative samples) may select customers who have not paid off the reduced amount. These customers may fail to complete the premium payment on time, resulting in changes in the policy status or incomplete insurance plans.

[0056] In this embodiment, the downsampling ratio can also be determined according to actual conditions. For example, while keeping the number of positive samples unchanged, the number of negative samples can be reduced to the order of c times (for example, 2 times, 3 times or 10 times, etc., which are not limited in this embodiment) the number of positive samples, and the dynamic random synthesis algorithm can be combined to complete oversampling of minority classes or undersampling of majority classes.

[0057] Optionally, in this embodiment, a dynamic random synthesis method can be used to generate synthetic samples, so that the differences between samples are preserved to ensure that the generated synthetic samples can effectively cover the entire data space, while avoiding the problem of overfitting.

[0058] In an optional implementation of this embodiment, if the target negative sample screening rule is an activity information screening rule, the data characteristics of the target activity can be determined, and a first negative sample whose correlation with the target activity is lower than a set threshold can be determined based on the data characteristics; further, the first negative sample can be determined as the target negative sample, and the target negative sample can be deleted; wherein the data characteristics include attendance data and activity data.

[0059] Optionally, in this embodiment, the data features of attendance data and activity data may be included first. For example, attendance data may include indicators such as attendance rate and working hours; activity data may include information such as participation in online and offline activities and activity frequency. Further, based on the determined activity data features, a first negative sample with a low correlation with the sales performance of the service expansion channel can be identified. For example, employees or customers with low attendance rates or low frequency of activity participation may be considered as negative samples. Further, it can be based on downsampling of attendance data: for attendance data, downsampling can be performed based on the attendance rate indicator. Employee data with an attendance rate less than the business threshold is selected as a negative sample, and the downsampling ratio is determined according to business needs to maintain a balance between positive and negative samples. Based on downsampling of activity data: for activity data, downsampling is performed based on participation in online and offline activities. Customer data with an activity participation frequency less than the business threshold is selected as a negative sample, and the downsampling ratio is determined according to actual conditions to maintain a balance between positive and negative samples.

[0060] In an optional implementation of the present embodiment, if the target negative sample screening rule is a phone appointment and face-to-face meeting screening rule, the target user who failed to make a phone appointment, or succeeded in making a phone appointment but failed to sign the contract can be determined, and the negative sample corresponding to the target user can be determined as the target negative sample; further, the target negative sample can be deleted.

[0061] Optionally, in this embodiment, for customers who failed to make appointments by phone, customers who failed to make appointments by phone can be identified by screening the appointment records. It is understandable that the failure of the appointment by phone may be due to the customer's lack of interest, time conflict or other reasons. These customers can be regarded as negative samples because they have no further opportunities for interviews and business transactions; for customers who successfully met after the appointment by phone but failed to issue the policy, the issuance result can be confirmed through business system records or feedback data. If the customer successfully met with the agent but ultimately failed to complete the policy issuance process, these customers can also be regarded as negative samples; this may be due to the customer changing his mind, not meeting the insurance requirements or other factors. In this embodiment, these customers can be used as target negative samples and can be deleted.

[0062] In a specific example of this embodiment, the serial path of phone appointments and face-to-face meetings can be used for stratified downsampling. Taking into account the sample feature distribution, the samples are divided into different levels according to the characteristic attributes of the samples, and downsampling is performed within each level. The stratified downsampling can be performed based on features such as whether the phone appointment is successful or not -> entering the face-to-face meeting stage -> whether the face-to-face meeting is successful or not. Within each subcategory, downsampling can be performed based on specific circumstances. For example, in the main category of whether the phone appointment is successful or not, if the number of samples of successful phone appointments is much greater than the number of samples of failed phone appointments, the samples of successful phone appointments can be downsampled to balance the ratio of positive and negative samples. Similarly, in the subcategory of whether the face-to-face meeting is successful or not, the samples can also be downsampled based on actual conditions to ensure that the sample distribution of each subcategory is balanced and to reduce the number of negative samples.

[0063] Step 140: Determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets a preset ratio requirement.

[0064] The first ratio is smaller than the second ratio.

[0065] In an optional implementation of the present embodiment, after obtaining the screened target sample set, the second ratio of the number of positive samples to the number of negative samples in the target sample set can be further determined, and it can be determined whether the second ratio meets the preset ratio requirement. If the second ratio meets the preset ratio requirement, the screening of target negative samples in the target sample set can be stopped, and subsequent model training can be performed based on the screened target sample set.

[0066] In another optional implementation of this embodiment, if the second ratio does not meet the preset ratio requirement, the target sample set may continue to be screened for negative samples to reduce the ratio of positive and negative samples in the target sample set.

[0067] Optionally, in this embodiment, after determining that the second ratio meets the preset ratio requirement, it may also include: inputting the screened target sample set into a target machine learning model, iteratively training the target machine learning model to obtain a target model; the target model is used to recommend new users who match the target insurance product.

[0068] In this embodiment, after screening the negative samples in each sample set, each sample set can be simultaneously input into the target machine learning model for iterative training. When the iteration stop condition is met, a model for recommending new users matching the target product is obtained through training.

[0069] Exemplarily, after training a model for recommending new users matching the target product, any insurance data may be input into the model to determine whether the user is a recommended user, that is, whether the user is more likely to purchase the target insurance product.

[0070] The technical solution of the embodiment of the present invention is to obtain an original insurance data set and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set; wherein the positive samples are insurance data of target insurance products that have been insured, and the negative samples are insurance data of target insurance products that have not been insured; when the first ratio does not meet the preset ratio requirement, determine each sample set corresponding to the original insurance data set, and determine the target negative sample screening rule that matches the target sample set; based on the target negative sample screening rule, screen each negative sample in the target sample set to obtain a screened target sample set; determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets the preset ratio requirement; the first ratio is less than the second ratio, which can reduce the number of negative samples in the training data and increase the ratio of positive and negative samples, and can provide a basis for training a recommendation model for high-quality potential insurance users.

[0071] Embodiment 2

[0072] Figure 2 is a flowchart of a negative sample screening method provided according to the second embodiment of the present invention. This embodiment is a further refinement of the above technical solution. The technical solution in this embodiment can be combined with each optional solution in one or more of the above embodiments. Figure 2 As shown, the method includes:

[0073] Step 210: Obtain an original insurance data set, and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set.

[0074] Step 220: When the first ratio does not meet the preset ratio requirement, determine each sample set corresponding to the original insurance data set, and determine a target negative sample screening rule that matches the target sample set.

[0075] In an optional implementation of this embodiment, determining each sample set corresponding to the original insurance data set may include: respectively determining the data category of each positive sample and each negative sample, and classifying each positive sample and each negative sample according to the data category to obtain multiple sample sets.

[0076] The data categories include: product data, service activity data or telephone appointment data, etc., which are not limited in this embodiment; wherein the data categories of each positive sample and each negative sample in each sample set are the same.

[0077] Optionally, in this embodiment, when it is determined that the first ratio of the number of positive samples to the number of negative samples in the original insurance data set does not meet the preset ratio, the data category of each positive sample and negative sample in the original insurance data set can be further determined separately, and each positive sample and negative sample can be classified according to the data category to obtain multiple sample sets; illustratively, positive samples and negative samples of the same data category can be divided into the same sample set.

[0078] Optionally, in this embodiment, determining the target negative sample screening rules that match the target sample set may include: determining the target negative sample screening rules that match the target sample set based on the data category of each positive sample or negative sample in the target sample set; the target negative sample screening rules include: product information screening rules, activity information screening rules, and telephone appointment and face-to-face screening rules.

[0079] It can be understood that if the data categories of the positive samples and negative samples in the same sample set are the same, then after the sample sets are divided, the target negative sample screening rules that match them can be determined according to the data category of each sample set; illustratively, if the data category of the target sample set is product data, then the target negative sample screening rules that match it are product information screening rules; if the data category of the target sample set is service activity data, then the target negative sample screening rules that match it are activity information screening rules; if the data category of the target sample set is telephone appointment face-to-face data, then the target negative sample screening rules that match it are telephone appointment face-to-face screening rules; that is, in this embodiment, the data categories and the target negative sample screening rules are matched one by one.

[0080] Step 230: Determine the target negative samples in the target sample set based on the target negative sample screening rule; delete the target negative samples to obtain the screened target sample set.

[0081] In an optional implementation of this embodiment, after determining the target negative sample screening rules that match the target sample set, the target negative samples in the target sample set can be further screened out based on the target negative sample screening rules, and the target negative samples can be deleted to obtain a screened target sample set.

[0082] Step 240: determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets a preset ratio requirement.

[0083] In order to better understand the negative sample screening method involved in this embodiment, Figure 3 is a flowchart of another negative sample screening method provided according to the second embodiment of the present invention, referring to Figure 3 , which mainly includes: obtaining original insurance data from the business system, and preprocessing the obtained original insurance data to obtain modeling data, which includes positive samples and negative samples; further, negative samples can be screened based on different screening rules. For example, three sample modules are obtained in the figure, each of which contains positive samples and negative samples, and the ratio of positive to negative samples is moderate; finally, these sample modules can be input into the machine learning model for iterative training to obtain the final insurance model suitable for new user recommendation.

[0084] The solution of this embodiment, after determining the target negative sample screening rules that match the target sample set, can further screen out the target negative samples in the target sample set based on the target negative sample screening rules, and delete the target negative samples to obtain a screened target sample set, thereby reducing the negative samples in the target sample set and providing a basis for improving the ratio of positive and negative samples in the target sample set.

[0085] Embodiment 3

[0086] Figure 4 Schematic diagram of a negative sample screening device according to Embodiment 3 of the present invention. Figure 4 As shown, the device includes: a first ratio determination module 410 , a sample set determination module 420 , a negative sample screening module 430 and a second ratio determination module 440 .

[0087] A first ratio determination module 410, configured to obtain an original insurance data set and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set;

[0088] The positive sample is the insurance data of the insured target insurance product, and the negative sample is the insurance data of the uninsured target insurance product;

[0089] A sample set determination module 420, for determining each sample set corresponding to the original insurance data set and determining a target negative sample screening rule matching the target sample set when the first ratio does not meet the preset ratio requirement;

[0090] A negative sample screening module 430 screens each negative sample in the target sample set based on the target negative sample screening rule to obtain a screened target sample set;

[0091] The second ratio determination module 440 is used to determine the second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets the preset ratio requirement; the first ratio is less than the second ratio.

[0092] The solution of this embodiment is to obtain the original insurance data set through a first ratio determination module, and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set; to determine each sample set corresponding to the original insurance data set through a sample set determination module when the first ratio does not meet the preset ratio requirement, and to determine a target negative sample screening rule that matches the target sample set; to screen each negative sample in the target sample set based on the target negative sample screening rule through a negative sample screening module to obtain a screened target sample set; to determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set through a second ratio determination module, and to stop screening the target sample set when the second ratio meets the preset ratio requirement; the first ratio is smaller than the second ratio, which can reduce the number of negative samples in the training data and improve the ratio of positive and negative samples, thereby providing a basis for training a recommendation model for high-quality potential insurance users.

[0093] In an optional implementation of this embodiment, the first ratio determination module 410 is specifically configured to obtain original insurance data from each target system respectively;

[0094] The original insurance data includes at least one of the following:

[0095] User information, policy information, claims information, and historical transaction records;

[0096] Preprocessing each of the original insurance data to obtain an original insurance data set;

[0097] Respectively determine the positive samples and negative samples in the original insurance data set, and count the number of the positive samples and the number of the negative samples;

[0098] The ratio of the number of the positive samples to the number of the negative samples is determined as the first ratio.

[0099] In an optional implementation of this embodiment, the sample set determination module 420 is specifically used to determine the data category of each of the positive samples and each of the negative samples, and classify each of the positive samples and the negative samples according to each of the data categories to obtain multiple sample sets;

[0100] The data categories include: product data, service activity data or telephone appointment data;

[0101] Among them, the data categories of each positive sample and each negative sample in each sample set are the same.

[0102] In an optional implementation of this embodiment, the sample set determination module 420 is further configured to determine a target negative sample screening rule matching the target sample set according to the data category of each positive sample or negative sample in the target sample set;

[0103] The target negative sample screening rules include: product information screening rules, activity information screening rules and phone appointment and face-to-face screening rules.

[0104] In an optional implementation of this embodiment, the target negative sample screening rule is determined by the following steps:

[0105] If the target negative sample screening rule is a product information screening rule, the target insurance customer in the target insurance product is determined, the target insurance customer is determined as a target negative sample, and the target negative sample is deleted; wherein the target insurance customer is a customer who fails to pay the premium on time and / or is rejected for insurance;

[0106] If the target negative sample screening rule is an activity information screening rule, then determine the data features of the target activity, and determine a first negative sample whose correlation with the target activity is lower than a set threshold according to the data features, determine the first negative sample as the target negative sample, and delete the target negative sample; the data features include attendance data and activity data;

[0107] If the target negative sample screening rule is a phone appointment and face-to-face meeting screening rule, the target user whose phone appointment failed, or whose phone appointment succeeded but the contract was not signed successfully, is determined, and the negative sample corresponding to the target user is determined as the target negative sample, and the target negative sample is deleted.

[0108] In an optional implementation of this embodiment, the negative sample screening module 430 is specifically configured to determine a target negative sample in the target sample set based on the target negative sample screening rule;

[0109] The target negative samples are deleted to obtain the filtered target sample set.

[0110] In an optional implementation of this embodiment, the negative sample screening device further includes: a training module, which is used to input the screened target sample set into a target machine learning model, and iteratively train the target machine learning model to obtain a target model;

[0111] The target model is used to recommend new users who match the target insurance product.

[0112] The negative sample screening device provided in the embodiment of the present invention can execute the negative sample screening method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0113] In the technical solutions of the embodiments of the present invention, the collection, storage, use, processing, transmission, provision and disclosure of user personal information (such as identity information, insurance information, etc.) are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0114] Embodiment 4

[0115] Figure 5 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0116] like Figure 5 As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0117] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0118] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 11 executes the various methods and processes described above, such as a method for screening negative samples.

[0119] In some embodiments, the negative sample screening method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the negative sample screening method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the negative sample screening method in any other appropriate manner (e.g., by means of firmware).

[0120] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0121] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0122] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0123] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0124] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0125] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0126] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0127] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for screening negative samples, characterized in that: include: Acquire an original insurance data set, and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set; wherein the positive samples are insurance data of insured target insurance products, and the negative samples are insurance data of uninsured target insurance products; In the case where the first ratio does not meet the preset ratio requirement, determining each sample set corresponding to the original insurance data set, and determining a target negative sample screening rule matching the target sample set; Screening each negative sample in the target sample set based on the target negative sample screening rule to obtain a screened target sample set; Determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets a preset ratio requirement; the first ratio is less than the second ratio.

2. The negative sample screening method according to claim 1, characterized in that: The obtaining of the original insurance data set and determining a first ratio of the number of positive samples to the number of negative samples in the original insurance data set includes: Obtain original insurance data from each target system respectively; The original insurance data includes at least one of the following: user information, policy information, claim information, and historical transaction records; Preprocessing each of the original insurance data to obtain an original insurance data set; Respectively determine the positive samples and negative samples in the original insurance data set, and count the number of the positive samples and the number of the negative samples; The ratio of the number of the positive samples to the number of the negative samples is determined as the first ratio.

3. The negative sample screening method according to claim 1, characterized in that: The determining of each sample set corresponding to the original insurance data set includes: Determining the data category of each of the positive samples and each of the negative samples respectively, and classifying each of the positive samples and the negative samples according to the data categories to obtain multiple sample sets; The data categories include: product data, service activity data or telephone appointment data; Among them, the data categories of each positive sample and each negative sample in each sample set are the same.

4. The method for screening negative samples according to claim 3, characterized in that: The step of determining a target negative sample screening rule that matches the target sample set includes: Determining a target negative sample screening rule that matches the target sample set according to the data category of each positive sample or negative sample in the target sample set; The target negative sample screening rules include: product information screening rules, activity information screening rules and phone appointment and face-to-face screening rules.

5. The method for screening negative samples according to claim 4, characterized in that: The target negative sample screening rule is determined by the following steps: If the target negative sample screening rule is a product information screening rule, the target insurance customer in the target insurance product is determined, the target insurance customer is determined as a target negative sample, and the target negative sample is deleted; wherein the target insurance customer is a customer who fails to pay the premium on time and / or is rejected for insurance; If the target negative sample screening rule is an activity information screening rule, then determine the data features of the target activity, and determine a first negative sample whose correlation with the target activity is lower than a set threshold according to the data features, determine the first negative sample as the target negative sample, and delete the target negative sample; the data features include attendance data and activity data; If the target negative sample screening rule is a phone appointment and face-to-face meeting screening rule, the target user whose phone appointment failed, or whose phone appointment succeeded but the contract was not signed successfully, is determined, and the negative sample corresponding to the target user is determined as the target negative sample, and the target negative sample is deleted.

6. The method for screening negative samples according to claim 5, characterized in that: The step of screening each negative sample in the target sample set based on the target negative sample screening rule to obtain a screened target sample set includes: Based on the target negative sample screening rule, determining the target negative sample in the target sample set; The target negative samples are deleted to obtain the filtered target sample set.

7. The method for screening negative samples according to claim 1, characterized in that: After determining that the second ratio meets the preset ratio requirement, the method further includes: Inputting the screened target sample set into a target machine learning model, and iteratively training the target machine learning model to obtain a target model; The target model is used to recommend new users who match the target insurance product.

8. A negative sample screening device, characterized in that: include: A first ratio determination module, configured to obtain an original insurance data set and determine a first ratio of the number of positive samples to the number of negative samples in the original insurance data set; The positive sample is the insurance data of the insured target insurance product, and the negative sample is the insurance data of the uninsured target insurance product; A sample set determination module, used for determining each sample set corresponding to the original insurance data set and determining a target negative sample screening rule matching the target sample set when the first ratio does not meet the preset ratio requirement; A negative sample screening module, which screens each negative sample in the target sample set based on the target negative sample screening rule to obtain a screened target sample set; A second ratio determination module is used to determine a second ratio of the number of positive samples to the number of negative samples in the screened target sample set, and stop screening the target sample set when the second ratio meets a preset ratio requirement; the first ratio is smaller than the second ratio.

9. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can perform the negative sample screening method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the negative sample screening method according to any one of claims 1 to 7 when executed.

Images