CN116993516A - Interpretation optimization method, device and equipment for claim settlement model and storage medium thereof - Google Patents
Interpretation optimization method, device and equipment for claim settlement model and storage medium thereof Download PDFInfo
- Publication number
- CN116993516A CN116993516A CN202311007444.3A CN202311007444A CN116993516A CN 116993516 A CN116993516 A CN 116993516A CN 202311007444 A CN202311007444 A CN 202311007444A CN 116993516 A CN116993516 A CN 116993516A
- Authority
- CN
- China
- Prior art keywords
- interpretation
- model
- sample
- data set
- settlement
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000005457 optimization Methods 0.000 title claims abstract description 23
- 238000012360 testing method Methods 0.000 claims description 219
- 238000012549 training Methods 0.000 claims description 108
- 238000004364 calculation method Methods 0.000 claims description 28
- 238000012545 processing Methods 0.000 claims description 20
- 238000000605 extraction Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 10
- 235000008733 Citrus aurantifolia Nutrition 0.000 claims description 7
- 235000011941 Tilia x europaea Nutrition 0.000 claims description 7
- 239000004571 lime Substances 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 abstract description 13
- 230000000694 effects Effects 0.000 description 7
- 238000013473 artificial intelligence Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 235000019800 disodium phosphate Nutrition 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003058 natural language processing Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/08—Insurance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Finance (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Accounting & Taxation (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Evolutionary Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Medical Informatics (AREA)
- Technology Law (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the application belongs to the technical field of financial science and technology, and relates to an interpretation and optimization method, a device, equipment and a storage medium of an insurance company claim settlement model, which are applied to an insurance company claim settlement model interpretation and optimization scene.
Description
Technical Field
The application relates to the technical field of finance and technology, and is applied to an insurance company claim model interpretation optimization scene, in particular to a method, a device and equipment for claim model interpretation optimization and a storage medium thereof.
Background
With the rapid development of the financial industry, more and more financial companies have more complicated business, particularly insurance claim risk prediction business and a good claim settlement model, not only can help enterprises avoid claim settlement risk, but also can enable clients to easily accept prediction results.
For claim model prediction, it is a common method to build a prediction model using neural network algorithms in enterprise insurance claim settlement. The neural network model has strong nonlinear modeling capability and can adapt to complex service scenes and data characteristics. However, deep learning means a black box test that is less interpretable and not easily understood and accepted by the customer.
Disclosure of Invention
The embodiment of the application aims to provide a method, a device, equipment and a storage medium for explaining and optimizing a claim model, so as to solve the problems that the explaining of the claim model in the prior art is low and is not easy to be understood and accepted by customers.
In order to solve the above technical problems, the embodiment of the present application provides an interpretation and optimization method for a claim model, which adopts the following technical scheme:
An interpretation optimization method of a claim model, comprising the following steps:
acquiring all the claim settlement business data of the closed case;
dividing the claim settlement business data into a training data set and a test data set according to a preset proportion;
inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a claim settlement model after preliminary training;
generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set respectively;
inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula;
and deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business.
Further, before executing the step of inputting the training data set into the pre-constructed claim settlement model to perform model training to obtain the claim settlement model after the preliminary training, the method further includes:
Identifying normal claim data and suspicious claim data in the training data set according to a preset claim type identifier, wherein the claim type identifier comprises the normal claim data identifier and the suspicious claim data identifier;
setting a normal claim data output node and an in-doubt claim data output node in the pre-constructed claim model;
a claim feature extraction component is preset in the pre-constructed claim model;
the step of inputting the training data set into a pre-constructed claim settlement model for model training to obtain a preliminary trained claim settlement model specifically comprises the following steps:
extracting all the claim characteristic information contained in each claim service data in the training data set respectively through the claim characteristic extraction component;
according to all the claim characteristic information contained in each claim service data, model training is carried out to obtain a model training result;
based on the model training result, counting the claim settlement business data respectively contained in the normal claim settlement data output node and the doubtful claim settlement data output node;
calculating the loss degree when model training is carried out according to the claim settlement business data respectively contained in the normal claim settlement data output node and the claim settlement data divided into the normal claim settlement data and the claim settlement data through a preset loss function;
If the loss degree does not meet the preset loss condition, adjusting the super parameters of the pre-constructed claim settlement model, and carrying out model training again;
and finishing the preliminary training of the pre-constructed claim settlement model until the loss degree meets a preset loss condition, and stopping model training.
Further, the preset model interpretation component includes a sample disturbance component based on a LIME algorithm, and the step of generating an interpretation data set corresponding to the test data set according to the preset model interpretation component specifically includes:
screening any sample to be interpreted from the test data set as a target test sample;
based on the feature extraction component, obtaining all claim feature information corresponding to each target test sample;
according to all the claim characteristic information corresponding to each target test sample, sample disturbance is respectively carried out on each target test sample by adopting a sample disturbance component based on a LIME algorithm, and all similar samples generated after each target test sample is subjected to sample disturbance are obtained;
and adding all similar samples generated after each target test sample is disturbed by the sample into a preset set to generate the interpretation data set.
Further, the step of performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set, specifically includes:
screening out similar samples corresponding to each target test sample from the interpretation data set;
calculating the editing distance between each similar sample and the corresponding target test sample by adopting an editing distance algorithm;
based on the editing distance between each similar sample and the corresponding target test sample, obtaining the average editing distance between each target test sample and all the similar samples corresponding to the target test sample;
according to the editing distance between each similar sample and the corresponding target test sample and the average editing distance between each target test sample and all the similar samples corresponding to the target test sample, acquiring the editing distance weight corresponding to each similar sample;
and respectively taking the editing distance weight corresponding to each similar sample as the interpretation weight of the corresponding sample in the interpretation data set.
Further, the step of obtaining the average edit distance between each target test sample and all the similar samples corresponding to each target test sample based on the edit distance between each similar sample and the corresponding target test sample specifically includes:
Selecting one target test sample from all target test samples as a current sample to be calculated;
carrying out weighted average processing on the editing distances between the current sample to be calculated and all corresponding similar samples by using a Gaussian kernel function to obtain a weighted average value;
taking the weighted average value as the average editing distance between the current sample to be calculated and all corresponding similar samples;
and sequentially taking different target test samples as current samples to be calculated, and acquiring average editing distances between each target test sample and all corresponding similar samples.
Further, the step of obtaining the edit distance weight corresponding to each similar sample according to the edit distance between each similar sample and the corresponding target test sample and the average edit distance between each target test sample and all the similar samples corresponding to each target test sample, specifically includes:
taking the average editing distance between each target test sample and all corresponding similar samples as a weight calculation hyper-parameter, and configuring the weight calculation hyper-parameter to a weight calculation model based on a neural network;
taking the edit distance between each similar sample and the corresponding target test sample as an input parameter, inputting the edit distance into the weight calculation model, and carrying out model operation by combining the weight calculation superparameter and a weight calculation formula preset in the weight calculation model, wherein the preset weight calculation formula is as follows: Wherein i represents the number information of each similar sample of the current target test sample, i is a positive integer, j represents the number information of the current target test sample in the test data set, j is a positive integer, ω i Representing edit distance weights for similar samples numbered i,representing an exponential-arithmetic function, +.>Represents the edit distance, alpha, between the similar sample numbered i and the current target test sample j Representing the current target test sample,/->Representing similar samples with the number i corresponding to the current target test sample, wherein sigma is obtained by calculating the average editing distance, and represents the discrete degree of the editing distance between the target test sample and all the similar samples corresponding to the target test sample relative to the average editing distance;
and obtaining an operation result of the weight calculation model, and obtaining the editing distance weight corresponding to each similar sample by analyzing the operation result.
Further, the step of inputting the test dataset, the interpretation dataset, and the interpretation weights corresponding to each sample in the interpretation dataset into the claim model after the preliminary training, and performing linear fitting on all samples in the interpretation dataset to obtain a model interpretation formula specifically includes:
Obtaining a model output value corresponding to each target test sample as a first predicted value according to the test data set and the claim settling model after preliminary training;
obtaining model output values respectively corresponding to each similar sample of each target test sample according to the interpretation data set and the initially trained claim settling model as a secondA predicted value; according to the first predicted value, the second predicted value, the interpretation weight corresponding to each sample in the interpretation data set and a preset linear fitting formula, performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula, wherein the preset linear fitting formula is as follows:wherein i represents the number information of each similar sample of the current target test sample, i is a positive integer, j represents the number information of the current target test sample in the test data set, j is a positive integer, ω i Interpretation weight, f (a), representing the corresponding similarity sample numbered i for the current target test sample j ) Representing said first predicted value, +.>Representing a second predicted value, a, corresponding to a similar sample numbered i corresponding to the current target test sample j Representing the current target test sample,/->And representing the similar sample with the number i corresponding to the current target test sample.
In order to solve the technical problems, the embodiment of the application also provides a claim model interpretation optimizing device, which adopts the following technical scheme:
an interpretation optimizing apparatus for claim model, comprising:
the claim settlement business data acquisition module is used for acquiring all the claim settlement business data of the established case;
the data set dividing module is used for dividing the claim settlement business data into a training data set and a test data set according to a preset proportion;
the claim model preliminary training module is used for inputting the training data set into a pre-constructed claim model, and carrying out model training to obtain a claim model with the preliminary training completed;
the sample interpretation weight acquisition module is used for generating an interpretation data set corresponding to the test data set according to a preset model interpretation assembly, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights respectively corresponding to each sample in the interpretation data set;
the model interpretation formula fitting module is used for inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training is completed, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula;
And the claim settlement model optimization module is used for deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business.
In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical schemes:
a computer device comprising a memory having stored therein computer readable instructions which when executed by the processor implement the steps of the claim model interpretation optimization method described above.
In order to solve the above technical problems, an embodiment of the present application further provides a computer readable storage medium, which adopts the following technical schemes:
a computer readable storage medium having stored thereon computer readable instructions which when executed by a processor implement the steps of the claim model interpretation optimization method as described above.
Compared with the prior art, the embodiment of the application has the following main beneficial effects:
the interpretation and optimization method of the claim model provided by the embodiment of the application is implemented by acquiring all the claim settlement business data of the established case; dividing the claim settlement business data into a training data set and a test data set according to a preset proportion; inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a claim settlement model after preliminary training; generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set respectively; inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula; and deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business. According to the method, sample disturbance is carried out on each target test sample according to a preset model interpretation component, all similar samples generated after sample disturbance on each target test sample are obtained, the effect of enriching sample quantity is achieved, the sample quantity for carrying out model test is increased, compared with test results obtained by a small number of test samples, more test results obtained by the test samples are more likely to be convinced by customers, the optimized claim settlement model is adopted for carrying out claim settlement prediction on financial business, the interpretation of the claim settlement model is improved, and the trust degree of customers on the claim settlement model is increased.
Drawings
In order to more clearly illustrate the solution of the present application, a brief description will be given below of the drawings required for the description of the embodiments of the present application, it being apparent that the drawings in the following description are some embodiments of the present application, and that other drawings may be obtained from these drawings without the exercise of inventive effort for a person of ordinary skill in the art.
FIG. 1 is an exemplary system architecture diagram in which the present application may be applied;
FIG. 2 is a flow chart illustrating one embodiment of an optimization method for a claims model according to the present application;
FIG. 3 is a flow chart of one embodiment of step 203 shown in FIG. 2;
FIG. 4 is a flow chart of one embodiment of step 204 shown in FIG. 2;
FIG. 5 is a flow chart of one embodiment of step 407 shown in FIG. 4;
FIG. 6 is a flow chart of one embodiment of step 408 shown in FIG. 4;
FIG. 7 is a schematic diagram of an embodiment of an interpretation optimizing apparatus according to the application;
FIG. 8 is a schematic structural view of one embodiment of a computer device according to the present application.
Detailed Description
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description of the applications herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having" and any variations thereof in the description of the application and the claims and the description of the drawings above are intended to cover a non-exclusive inclusion. The terms first, second and the like in the description and in the claims or in the above-described figures, are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order.
Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the application. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.
In order to make the person skilled in the art better understand the solution of the present application, the technical solution of the embodiment of the present application will be clearly and completely described below with reference to the accompanying drawings.
As shown in fig. 1, a system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 is used as a medium to provide communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others.
The user may interact with the server 105 via the network 104 using the terminal devices 101, 102, 103 to receive or send messages or the like. Various communication client applications, such as a web browser application, a shopping class application, a search class application, an instant messaging tool, a mailbox client, social platform software, etc., may be installed on the terminal devices 101, 102, 103.
The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smartphones, tablet computers, electronic book readers, MP3 players (Moving Picture ExpertsGroup Audio Layer III, dynamic video expert compression standard audio plane 3), MP4 (Moving PictureExperts Group Audio Layer IV, dynamic video expert compression standard audio plane 4) players, laptop and desktop computers, and the like.
The server 105 may be a server providing various services, such as a background server providing support for pages displayed on the terminal devices 101, 102, 103.
It should be noted that, the method for optimizing interpretation of claim models provided by the embodiments of the present application is generally executed by a server/terminal device, and accordingly, the apparatus for optimizing interpretation of claim models is generally disposed in the server/terminal device.
It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
With continued reference to FIG. 2, a flow chart of one embodiment of a method of claim model interpretation optimization in accordance with the present application is shown. The method for explaining and optimizing the claim model comprises the following steps:
Step 201, obtaining all the claim settlement business data of the settled case.
Specifically, the claim settlement business data may be car insurance claim settlement business data, life insurance claim settlement business data, disease insurance claim settlement business data, etc.
And 202, dividing the claim settlement business data into a training data set and a test data set according to a preset proportion.
In this embodiment, the claim service data is divided into a training data set and a test data set according to a preset proportion, and specifically, a random sampling method may be adopted to obtain the training data set and the test data set of the preset proportion from the claim service data.
And 203, inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a preliminary trained claim settlement model.
In this embodiment, before executing the step of inputting the training data set into a pre-constructed claim settlement model to perform model training to obtain a claim settlement model after preliminary training, the method further includes: identifying normal claim data and suspicious claim data in the training data set according to a preset claim type identifier, wherein the claim type identifier comprises the normal claim data identifier and the suspicious claim data identifier; setting a normal claim data output node and an in-doubt claim data output node in the pre-constructed claim model; and presetting a claim feature extraction component in the pre-constructed claim model.
The normal claim data and the suspicious claim data in the training data set are identified according to the preset claim type identifier, wherein the claim type identifier comprises the normal claim data identifier and the suspicious claim data identifier, and specifically, the normal claim data and the suspicious claim data in the training data set can be determined according to whether all the claim service data of the deposited claim are finally successfully subjected to claim settlement.
The normal claim data output node and the suspicious claim data output node are arranged in the pre-constructed claim model, so that the normal claim data and the suspicious claim data in the claim service data can be predicted, risk avoidance of insurance companies is facilitated, and in addition, an artificial intelligence mode is adopted to replace manual risk prediction, so that the risk prediction is more intelligent and automatic.
The method comprises the steps of setting an claim characteristic extraction component in the pre-built claim model in advance, firstly extracting claim characteristic information from a data set input into the claim model during training or prediction, carrying out model training or prediction according to the extracted claim characteristic information, specifically, the claim characteristic extraction component can be formed by adopting a supervised neural learning network architecture, if the claim characteristic information is required to be imported in advance, analyzing claim service data according to the claim characteristic label, obtaining claim characteristic information, or adopting an unsupervised neural learning network architecture, if the claim characteristic information is required to be clustered according to a clustering algorithm, and determining the number of the claim characteristic information according to a clustering result and the number of final clustering clusters, so as to obtain corresponding claim characteristic information.
With continued reference to fig. 3, fig. 3 is a flow chart of one embodiment of step 203 shown in fig. 2, comprising:
step 301, extracting all the claim feature information contained in each piece of claim business data in the training data set through the claim feature extraction component;
step 302, performing model training according to all claim characteristic information contained in each claim service data respectively to obtain a model training result;
step 303, based on the model training result, counting the claim settlement business data respectively contained in the normal claim settlement data output node and the doubtful claim settlement data output node;
step 304, calculating the loss degree during model training according to the claim settlement business data respectively contained in the normal claim settlement data output node and the claim settlement data divided into the normal claim settlement data and the claim settlement data through a preset loss function;
specifically, the preset loss function is used for representing differentiation between the output result and the expected result.
Step 305, if the loss degree does not meet the preset loss condition, adjusting the super parameters of the pre-constructed claim model, and re-performing model training;
And 306, finishing the preliminary training of the pre-constructed claim settlement model until the loss degree meets a preset loss condition, and stopping model training.
In this embodiment, the preset loss condition specifically refers to whether the loss degree is smaller than a preset loss degree threshold, if so, the preset loss condition is satisfied, and if not, the preset loss condition is not satisfied.
And 204, generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set.
In this embodiment, the preset model interpretation component includes a sample perturbation component based on the LIME algorithm.
With continued reference to fig. 4, fig. 4 is a flow chart of one embodiment of step 204 shown in fig. 2, comprising:
step 401, screening out any sample to be interpreted from the test data set as a target test sample;
step 402, obtaining all claim feature information corresponding to each target test sample based on the feature extraction component;
step 403, according to all the claim feature information corresponding to each target test sample, sample disturbance is respectively carried out on each target test sample by adopting a sample disturbance component based on a LIME algorithm, and all similar samples generated after each target test sample is subjected to sample disturbance are obtained;
Sample disturbance is respectively carried out on each target test sample by adopting a sample disturbance component based on the LIME algorithm, all similar samples generated after each target test sample is subjected to sample disturbance are obtained, the effect of enriching the sample quantity is achieved, the test samples are taken as original samples, similar samples are generated, the sample quantity for carrying out model test is increased, and compared with test results obtained by a small number of test samples, more test results obtained by the test samples are more easily convinced by customers.
Step 404, adding all similar samples generated after each target test sample is disturbed by the sample into a preset set to generate the interpretation data set;
step 405, screening out similar samples corresponding to each target test sample from the interpretation data set;
step 406, calculating the editing distance between each similar sample and the corresponding target test sample by adopting an editing distance algorithm;
step 407, obtaining average editing distances between all similar samples corresponding to each target test sample based on the editing distances between each similar sample and the corresponding target test sample;
with continued reference to fig. 5, fig. 5 is a flow chart of one embodiment of step 407 shown in fig. 4, comprising:
Step 501, selecting one target test sample from all target test samples as a current sample to be calculated;
step 502, performing weighted average processing on the edit distances between the current sample to be calculated and all corresponding similar samples by using a Gaussian kernel function to obtain a weighted average;
step 503, taking the weighted average value as an average editing distance between the current sample to be calculated and all corresponding similar samples;
and 504, sequentially taking different target test samples as current samples to be calculated, and acquiring average editing distances between each target test sample and all corresponding similar samples.
Step 408, according to the edit distance between each similar sample and the corresponding target test sample and the average edit distance between each target test sample and all the similar samples corresponding to each target test sample, obtaining the edit distance weight corresponding to each similar sample;
with continued reference to FIG. 6, FIG. 6 is a flow chart of one embodiment of step 408 shown in FIG. 4, comprising:
step 601, taking the average editing distance between each target test sample and all corresponding similar samples as a weight calculation hyper-parameter, and configuring the weight calculation hyper-parameter to a weight calculation model based on a neural network;
Step 602, inputting the edit distance between each similar sample and the corresponding target test sample as an input parameter to the weight calculation model, and performing model operation by combining the weight calculation superparameter and a weight calculation formula preset in the weight calculation model, wherein the preset weight calculation formula is as follows:wherein i represents the number information of each similar sample of the current target test sample, i is a positive integer, j represents the number information of the current target test sample in the test data set, j is a positive integer, ω i Edit distance weight representing a similar sample numbered i, +.>Representing an exponential-arithmetic function, +.>Represents the edit distance, a, between the similar sample numbered i and the current target test sample j Representing the current target test sample,/->Representing similar samples with the number i corresponding to the current target test sample, wherein sigma is obtained by calculating the average editing distance, and represents the discrete degree of the editing distance between the target test sample and all the similar samples corresponding to the target test sample relative to the average editing distance;
and 603, obtaining an operation result of the weight calculation model, and obtaining the editing distance weight corresponding to each similar sample by analyzing the operation result.
In essence, the edit distance weight corresponding to each similar sample refers to the similarity of each similar sample relative to the target test sample.
And 409, taking the edit distance weight corresponding to each similar sample as the interpretation weight of the corresponding sample in the interpretation dataset.
And 205, inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula.
In this embodiment, the step of inputting the test dataset, the interpretation dataset, and the interpretation weights corresponding to each sample in the interpretation dataset into the claim model after the preliminary training, and performing linear fitting on all samples in the interpretation dataset to obtain a model interpretation formula specifically includes: obtaining a model output value corresponding to each target test sample as a first predicted value according to the test data set and the claim settling model after preliminary training; obtaining model output values corresponding to each similar sample of each target test sample respectively according to the interpretation data set and the preliminary training-completed claim settlement model, and taking the model output values as second predicted values; according to the first predicted value, the second predicted value, the interpretation weight corresponding to each sample in the interpretation data set and a preset linear fitting formula, performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula, wherein the preset linear fitting formula is as follows: Wherein i represents the number information of each similar sample of the current target test sample, i is a positive integer, j represents the number information of the current target test sample in the test data set, j is a positive integer, ω i Interpretation weight, f (a), representing the corresponding similarity sample numbered i for the current target test sample j ) Representing said first predicted value, +.>Representing a second predicted value, a, corresponding to a similar sample numbered i corresponding to the current target test sample j Representing the current target test sample,/->And representing the similar sample with the number i corresponding to the current target test sample.
And 206, deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business.
And deploying the model interpretation formula into the initially trained claim settlement model to generate an optimized claim settlement model, and adopting the optimized claim settlement model to carry out claim settlement prediction on financial business, so that the interpretation of the claim settlement model is improved, and the trust degree of clients on the claim settlement model is increased.
The application obtains all the business data of claims of the settled case; dividing the claim settlement business data into a training data set and a test data set according to a preset proportion; inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a claim settlement model after preliminary training; generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set respectively; inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula; and deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business. According to the method, sample disturbance is carried out on each target test sample according to a preset model interpretation component, all similar samples generated after sample disturbance on each target test sample are obtained, the effect of enriching sample quantity is achieved, the sample quantity for carrying out model test is increased, compared with test results obtained by a small number of test samples, more test results obtained by the test samples are more likely to be convinced by customers, the optimized claim settlement model is adopted for carrying out claim settlement prediction on financial business, the interpretation of the claim settlement model is improved, and the trust degree of customers on the claim settlement model is increased.
The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Among these, artificial intelligence (Artificial Intelligence, AI) is the theory, method, technique and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend and extend human intelligence, sense the environment, acquire knowledge and use knowledge to obtain optimal results.
Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.
In the embodiment of the application, sample disturbance is respectively carried out on each target test sample according to the preset model interpretation component, all similar samples generated after each target test sample is subjected to sample disturbance are obtained, the effect of enriching sample quantity is achieved, the sample quantity for carrying out model test is increased, compared with the test results obtained by a small number of test samples, the test results obtained by more test samples are easier to trust for customers, the optimized claim settlement model is adopted for carrying out claim settlement prediction on financial business, the interpretation of the claim settlement model is improved, and the trust degree of customers on the claim settlement model is increased.
With further reference to fig. 7, as an implementation of the method shown in fig. 2 described above, the present application provides an embodiment of a claim model interpretation optimizing apparatus, which corresponds to the method embodiment shown in fig. 2, and which is particularly applicable to various electronic devices.
As shown in fig. 7, the claim model interpretation optimizing apparatus 700 according to the present embodiment includes: the system comprises a claim settlement business data acquisition module 701, a data set division module 702, a claim settlement model preliminary training module 703, a sample interpretation weight acquisition module 704, a model interpretation formula fitting module 705 and a claim settlement model optimization module 706. Wherein:
the claim service data acquisition module 701 is configured to acquire all claim service data of the closed case;
the data set dividing module 702 is configured to divide the claim settlement business data into a training data set and a test data set according to a preset proportion;
the claim model preliminary training module 703 is configured to input the training data set into a pre-constructed claim model, perform model training, and obtain a claim model after the preliminary training is completed;
the sample interpretation weight obtaining module 704 is configured to generate an interpretation dataset corresponding to the test dataset according to a preset model interpretation component, and interpret all samples in the interpretation dataset to obtain interpretation weights corresponding to each sample in the interpretation dataset;
The model interpretation formula fitting module 705 is configured to input the test dataset, the interpretation dataset, and interpretation weights corresponding to each sample in the interpretation dataset into the claim model after the preliminary training is completed, and perform linear fitting on all samples in the interpretation dataset to obtain a model interpretation formula;
and the claim model optimizing module 706 is configured to deploy the model interpretation formula into the initially trained claim model, generate an optimized claim model, and perform claim settlement prediction on the financial service by using the optimized claim model.
The application obtains all the business data of claims of the settled case; dividing the claim settlement business data into a training data set and a test data set according to a preset proportion; inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a claim settlement model after preliminary training; generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set respectively; inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula; and deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business. According to the method, sample disturbance is carried out on each target test sample according to a preset model interpretation component, all similar samples generated after sample disturbance on each target test sample are obtained, the effect of enriching sample quantity is achieved, the sample quantity for carrying out model test is increased, compared with test results obtained by a small number of test samples, more test results obtained by the test samples are more likely to be convinced by customers, the optimized claim settlement model is adopted for carrying out claim settlement prediction on financial business, the interpretation of the claim settlement model is improved, and the trust degree of customers on the claim settlement model is increased.
Those skilled in the art will appreciate that implementing all or part of the above described embodiment methods may be accomplished by computer readable instructions, stored on a computer readable storage medium, that the program when executed may comprise the steps of embodiments of the methods described above. The storage medium may be a nonvolatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a random access Memory (Random Access Memory, RAM).
It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein. Moreover, at least some of the steps in the flowcharts of the figures may include a plurality of sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, the order of their execution not necessarily being sequential, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.
In order to solve the technical problems, the embodiment of the application also provides computer equipment. Referring specifically to fig. 8, fig. 8 is a basic structural block diagram of a computer device according to the present embodiment.
The computer device 8 comprises a memory 8a, a processor 8b, a network interface 8c communicatively connected to each other via a system bus. It should be noted that only computer device 8 having components 8a-8c is shown in the figures, but it should be understood that not all of the illustrated components need be implemented, and that more or fewer components may alternatively be implemented. It will be appreciated by those skilled in the art that the computer device herein is a device capable of automatically performing numerical calculations and/or information processing in accordance with predetermined or stored instructions, the hardware of which includes, but is not limited to, microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASICs), programmable gate arrays (fields-Programmable Gate Array, FPGAs), digital processors (Digital Signal Processor, DSPs), embedded devices, etc.
The computer equipment can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing equipment. The computer equipment can perform man-machine interaction with a user through a keyboard, a mouse, a remote controller, a touch pad or voice control equipment and the like.
The memory 8a includes at least one type of readable storage medium including flash memory, hard disk, multimedia card, card memory (e.g., SD or DX memory, etc.), random Access Memory (RAM), static Random Access Memory (SRAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), programmable Read Only Memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the storage 8a may be an internal storage unit of the computer device 8, such as a hard disk or a memory of the computer device 8. In other embodiments, the memory 8a may also be an external storage device of the computer device 8, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash Card (Flash Card) or the like, which are provided on the computer device 8. Of course, the memory 8a may also comprise both an internal memory unit of the computer device 8 and an external memory device. In this embodiment, the memory 8a is typically used to store an operating system and various application software installed on the computer device 8, such as computer readable instructions for interpreting an optimization method by a claim model. Further, the memory 8a may be used to temporarily store various types of data that have been output or are to be output.
The processor 8b may be a central processing unit (Central Processing Unit, CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments. The processor 8b is typically used to control the overall operation of the computer device 8. In this embodiment, the processor 8b is configured to execute computer readable instructions stored in the memory 8a or process data, such as computer readable instructions for executing the claim model interpretation optimization method.
The network interface 8c may comprise a wireless network interface or a wired network interface, which network interface 8c is typically used to establish a communication connection between the computer device 8 and other electronic devices.
The computer equipment provided by the embodiment belongs to the technical field of financial science and technology, and is applied to an insurance company claim settlement model interpretation optimization scene. The application obtains all the business data of claims of the settled case; dividing the claim settlement business data into a training data set and a test data set according to a preset proportion; inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a claim settlement model after preliminary training; generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set respectively; inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula; and deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business. According to the method, sample disturbance is carried out on each target test sample according to a preset model interpretation component, all similar samples generated after sample disturbance on each target test sample are obtained, the effect of enriching sample quantity is achieved, the sample quantity for carrying out model test is increased, compared with test results obtained by a small number of test samples, more test results obtained by the test samples are more likely to be convinced by customers, the optimized claim settlement model is adopted for carrying out claim settlement prediction on financial business, the interpretation of the claim settlement model is improved, and the trust degree of customers on the claim settlement model is increased.
The present application also provides another embodiment, namely, a computer-readable storage medium storing computer-readable instructions executable by a processor to cause the processor to perform the steps of the claim model interpretation optimization method as described above.
The computer readable storage medium provided by the embodiment belongs to the technical field of financial science and technology, and is applied to an insurance company claim settlement model interpretation optimization scene. The application obtains all the business data of claims of the settled case; dividing the claim settlement business data into a training data set and a test data set according to a preset proportion; inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a claim settlement model after preliminary training; generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set respectively; inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula; and deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business. According to the method, sample disturbance is carried out on each target test sample according to a preset model interpretation component, all similar samples generated after sample disturbance on each target test sample are obtained, the effect of enriching sample quantity is achieved, the sample quantity for carrying out model test is increased, compared with test results obtained by a small number of test samples, more test results obtained by the test samples are more likely to be convinced by customers, the optimized claim settlement model is adopted for carrying out claim settlement prediction on financial business, the interpretation of the claim settlement model is improved, and the trust degree of customers on the claim settlement model is increased.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present application.
It is apparent that the above-described embodiments are only some embodiments of the present application, but not all embodiments, and the preferred embodiments of the present application are shown in the drawings, which do not limit the scope of the patent claims. This application may be embodied in many different forms, but rather, embodiments are provided in order to provide a thorough and complete understanding of the present disclosure. Although the application has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments described in the foregoing description, or equivalents may be substituted for elements thereof. All equivalent structures made by the content of the specification and the drawings of the application are directly or indirectly applied to other related technical fields, and are also within the scope of the application.
Claims (10)
1. A method for optimizing interpretation of a claim model, comprising the steps of:
acquiring all the claim settlement business data of the closed case;
dividing the claim settlement business data into a training data set and a test data set according to a preset proportion;
inputting the training data set into a pre-constructed claim settlement model, and performing model training to obtain a claim settlement model after preliminary training;
generating an interpretation data set corresponding to the test data set according to a preset model interpretation component, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights corresponding to each sample in the interpretation data set respectively;
inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula;
and deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business.
2. The method of claim 1, wherein prior to performing the step of inputting the training dataset into a pre-constructed claim model for model training to obtain a preliminary trained claim model, the method further comprises:
Identifying normal claim data and suspicious claim data in the training data set according to a preset claim type identifier, wherein the claim type identifier comprises the normal claim data identifier and the suspicious claim data identifier;
setting a normal claim data output node and an in-doubt claim data output node in the pre-constructed claim model;
a claim feature extraction component is preset in the pre-constructed claim model;
the step of inputting the training data set into a pre-constructed claim settlement model for model training to obtain a preliminary trained claim settlement model specifically comprises the following steps:
extracting all the claim characteristic information contained in each claim service data in the training data set respectively through the claim characteristic extraction component;
according to all the claim characteristic information contained in each claim service data, model training is carried out to obtain a model training result;
based on the model training result, counting the claim settlement business data respectively contained in the normal claim settlement data output node and the doubtful claim settlement data output node;
calculating the loss degree when model training is carried out according to the claim settlement business data respectively contained in the normal claim settlement data output node and the claim settlement data divided into the normal claim settlement data and the claim settlement data through a preset loss function;
If the loss degree does not meet the preset loss condition, adjusting the super parameters of the pre-constructed claim settlement model, and carrying out model training again;
and finishing the preliminary training of the pre-constructed claim settlement model until the loss degree meets a preset loss condition, and stopping model training.
3. The method for optimizing interpretation of claim 1, wherein the preset interpretation component includes a sample perturbation component based on LIME algorithm, and the step of generating the interpretation dataset corresponding to the test dataset according to the preset interpretation component specifically includes:
screening any sample to be interpreted from the test data set as a target test sample;
based on the feature extraction component, obtaining all claim feature information corresponding to each target test sample;
according to all the claim characteristic information corresponding to each target test sample, sample disturbance is respectively carried out on each target test sample by adopting a sample disturbance component based on a LIME algorithm, and all similar samples generated after each target test sample is subjected to sample disturbance are obtained;
and adding all similar samples generated after each target test sample is disturbed by the sample into a preset set to generate the interpretation data set.
4. The method for optimizing interpretation model according to claim 3, wherein the step of performing interpretation processing on all samples in the interpretation dataset to obtain respective interpretation weights for each sample in the interpretation dataset comprises:
screening out similar samples corresponding to each target test sample from the interpretation data set;
calculating the editing distance between each similar sample and the corresponding target test sample by adopting an editing distance algorithm;
based on the editing distance between each similar sample and the corresponding target test sample, obtaining the average editing distance between each target test sample and all the similar samples corresponding to the target test sample;
according to the editing distance between each similar sample and the corresponding target test sample and the average editing distance between each target test sample and all the similar samples corresponding to the target test sample, acquiring the editing distance weight corresponding to each similar sample;
and respectively taking the editing distance weight corresponding to each similar sample as the interpretation weight of the corresponding sample in the interpretation data set.
5. The method for optimizing interpretation model according to claim 4, wherein the step of obtaining average edit distances between all the similar samples corresponding to each target test sample based on edit distances between each similar sample and the corresponding target test sample, comprises:
Selecting one target test sample from all target test samples as a current sample to be calculated;
carrying out weighted average processing on the editing distances between the current sample to be calculated and all corresponding similar samples by using a Gaussian kernel function to obtain a weighted average value;
taking the weighted average value as the average editing distance between the current sample to be calculated and all corresponding similar samples;
and sequentially taking different target test samples as current samples to be calculated, and acquiring average editing distances between each target test sample and all corresponding similar samples.
6. The method for optimizing interpretation model according to claim 4, wherein the step of obtaining the edit distance weight corresponding to each similar sample according to the edit distance between each similar sample and the corresponding target test sample and the average edit distance between each target test sample and all the similar samples corresponding to each target test sample, respectively, specifically comprises:
taking the average editing distance between each target test sample and all corresponding similar samples as a weight calculation hyper-parameter, and configuring the weight calculation hyper-parameter to a weight calculation model based on a neural network;
Taking the edit distance between each similar sample and the corresponding target test sample as an input parameter, inputting the edit distance into the weight calculation model, and carrying out model operation by combining the weight calculation superparameter and a weight calculation formula preset in the weight calculation model, wherein the preset weight calculation formula is as follows:wherein i represents the number information of each similar sample of the current target test sample, i is a positive integer, j represents the number information of the current target test sample in the test data set, j is a positive integer, ω i Representing edit distance weights for similar samples numbered i,representing an exponential-arithmetic function, +.>Represents the edit distance, a, between the similar sample numbered i and the current target test sample j Representing the current target test sample,/->Representing similar samples with the number i corresponding to the current target test sample, wherein sigma is obtained by calculating the average editing distance, and represents the discrete degree of the editing distance between the target test sample and all the similar samples corresponding to the target test sample relative to the average editing distance;
and obtaining an operation result of the weight calculation model, and obtaining the editing distance weight corresponding to each similar sample by analyzing the operation result.
7. The method for optimizing interpretation model interpretation of claim 6, wherein the steps of inputting the interpretation weights corresponding to each sample in the test dataset, the interpretation dataset, and the interpretation dataset into the initially trained claim model, and linearly fitting all samples in the interpretation dataset to obtain a model interpretation formula specifically include:
obtaining a model output value corresponding to each target test sample as a first predicted value according to the test data set and the claim settling model after preliminary training;
obtaining model output values corresponding to each similar sample of each target test sample respectively according to the interpretation data set and the preliminary training-completed claim settlement model, and taking the model output values as second predicted values;
according to the first predicted value, the second predicted value, the interpretation weight corresponding to each sample in the interpretation data set and a preset linear fitting formula, performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula, wherein the preset linear fitting formula is as follows:wherein i represents the number information of each similar sample of the current target test sample, i is a positive integer, j represents the number information of the current target test sample in the test data set, j is a positive integer, ω i Interpretation weight, f (a), representing the corresponding similarity sample numbered i for the current target test sample j ) Representing said first predicted value, +.>Representing a second predicted value, a, corresponding to a similar sample numbered i corresponding to the current target test sample j Representing the current target test sample,/->And representing the similar sample with the number i corresponding to the current target test sample.
8. An interpretation optimizing apparatus for claim model, comprising:
the claim settlement business data acquisition module is used for acquiring all the claim settlement business data of the established case;
the data set dividing module is used for dividing the claim settlement business data into a training data set and a test data set according to a preset proportion;
the claim model preliminary training module is used for inputting the training data set into a pre-constructed claim model, and carrying out model training to obtain a claim model with the preliminary training completed;
the sample interpretation weight acquisition module is used for generating an interpretation data set corresponding to the test data set according to a preset model interpretation assembly, and performing interpretation processing on all samples in the interpretation data set to obtain interpretation weights respectively corresponding to each sample in the interpretation data set;
the model interpretation formula fitting module is used for inputting the test data set, the interpretation data set and the interpretation weights corresponding to each sample in the interpretation data set into the claim model after preliminary training is completed, and performing linear fitting on all samples in the interpretation data set to obtain a model interpretation formula;
And the claim settlement model optimization module is used for deploying the model interpretation formula into the initially trained claim settlement model, generating an optimized claim settlement model, and adopting the optimized claim settlement model to perform claim settlement prediction on financial business.
9. A computer device comprising a memory having stored therein computer readable instructions which when executed by the processor implement the steps of the claim model interpretation optimization method of any of claims 1 to 7.
10. A computer readable storage medium having stored thereon computer readable instructions which when executed by a processor implement the steps of the claim model interpretation optimization method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311007444.3A CN116993516A (en) | 2023-08-10 | 2023-08-10 | Interpretation optimization method, device and equipment for claim settlement model and storage medium thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311007444.3A CN116993516A (en) | 2023-08-10 | 2023-08-10 | Interpretation optimization method, device and equipment for claim settlement model and storage medium thereof |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116993516A true CN116993516A (en) | 2023-11-03 |
Family
ID=88531856
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311007444.3A Pending CN116993516A (en) | 2023-08-10 | 2023-08-10 | Interpretation optimization method, device and equipment for claim settlement model and storage medium thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116993516A (en) |
-
2023
- 2023-08-10 CN CN202311007444.3A patent/CN116993516A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117395253A (en) | Load balancing method, device, equipment and storage medium thereof | |
| CN118656495B (en) | Public opinion publishing traceability method, device, equipment and storage medium thereof | |
| CN114513578A (en) | Outbound method, device, computer equipment and storage medium | |
| CN116644298A (en) | Method for detecting performance of network attack detection model and related equipment thereof | |
| CN114241411B (en) | Counting model processing method and device based on target detection and computer equipment | |
| CN117493563A (en) | Session intention analysis method, device, equipment and storage medium thereof | |
| CN116340864B (en) | Model drift detection method, device, equipment and storage medium thereof | |
| CN115238805B (en) | Training method of abnormal data recognition model and related equipment | |
| CN117235633A (en) | Mechanism classification method, mechanism classification device, computer equipment and storage medium | |
| CN116934283A (en) | Employee authority configuration method, device, equipment and storage medium thereof | |
| CN111091198A (en) | Data processing method and device | |
| CN116777646A (en) | Artificial intelligence-based risk identification method, apparatus, device and storage medium | |
| CN116933869A (en) | Map construction method, device, computer equipment and storage medium | |
| CN116703466A (en) | System access quantity prediction method based on improved wolf algorithm and related equipment thereof | |
| CN116450943A (en) | Artificial intelligence-based speaking recommendation method, device, equipment and storage medium | |
| CN116993516A (en) | Interpretation optimization method, device and equipment for claim settlement model and storage medium thereof | |
| CN115826973A (en) | List page generation method and device, computer equipment and storage medium | |
| CN117541868B (en) | Training method for image classification model, image classification method, model, computer device, and medium | |
| CN114925202B (en) | Text classification method, device, computer equipment and storage medium | |
| CN116777645A (en) | Method for enhancing and optimizing risk prediction model of vehicle risk and related equipment thereof | |
| CN117114894A (en) | Method, device and equipment for predicting claim result and storage medium thereof | |
| CN117034114A (en) | Data prediction method, device, equipment and storage medium based on artificial intelligence | |
| CN116824600A (en) | Company seal identification method and related equipment thereof | |
| CN119206809A (en) | Face recognition method, device, computer equipment and medium based on artificial intelligence | |
| CN119515399A (en) | A method, device, computer equipment and storage medium for recovering lost customers |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |