Flood insurance loss risk assessment method and device based on data driving

Abstract

The invention discloses a flood insurance loss risk assessment method and device based on data driving, and the method comprises the steps: constructing a disaster database which comprises a plurality of key features affecting the flood risk; key features influencing flood risks are screened, and a multi-dimensional feature data set is constructed; and constructing a plurality of evaluation models, training through the multi-dimensional feature data set, performing performance evaluation and interpretive analysis to obtain an evaluation result, selecting a final evaluation model, and performing flood loss risk evaluation. Through the method and the model algorithm provided by the invention, incomplete or unstructured data is effectively processed, the flexibility and adaptability of evaluation are improved, the credibility analysis of the model increases the trust of a decision maker to a prediction result, the effectiveness of the model in practical application is promoted, and the method is suitable for popularization and application. Therefore, the generalization ability and the prediction accuracy of the model are improved, and the flood risk can be evaluated and managed more scientifically and accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of flood disaster risk assessment, and in particular to a data-driven flood insurance loss risk assessment method and device. Background Art

[0002] At present, extreme weather events caused by global climate change, especially flood disasters, pose a threat to economic and social stability. Flood disasters have always seriously threatened the safety of people's lives and property and social stability and development. In recent years, extreme weather events have occurred frequently, and the frequency of rainstorms and floods and the losses caused by them have continued to increase. Therefore, rapid and scientific assessment of flood losses is an important basis for formulating disaster prevention, disaster resistance, disaster relief and post-disaster reconstruction plans, and is of great significance to flood prevention and disaster reduction. Flood loss assessment is a basic research work in the field of flood prevention and disaster reduction research. Conducting flood disaster loss assessment can provide important support for flood impact assessment, flood control benefit evaluation, preparation of flood emergency plans, rationality analysis of flood control planning, and formulation of flood prevention and disaster reduction countermeasures.

[0003] China faces severe challenges from flood disasters, but traditional flood risk assessment methods have problems such as high data dependence, complex models, high computational costs, and parameter uncertainty. Summary of the invention

[0004] To this end, the present invention aims to at least to some extent address the deficiencies in the prior art, thereby proposing a data-driven flood insurance loss risk assessment method and device.

[0005] In a first aspect, the present invention provides a data-driven flood insurance assessment method, the method comprising:

[0006] Building a disaster database, wherein the disaster database includes a plurality of key features affecting flood risk;

[0007] Screening the key features that affect flood risk, and constructing a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socio-economic data, and actual flood loss data;

[0008] Constructing multiple evaluation models and training them through the multidimensional feature data set, performing performance evaluation and explanatory analysis on the multiple evaluation models after training to obtain evaluation results;

[0009] A final assessment model is selected based on the assessment results, and flood loss risk assessment is performed using the final assessment model.

[0010] In a second aspect, the present invention provides a data-driven flood insurance loss risk assessment system, the system comprising:

[0011] A construction module: used for constructing a disaster database, wherein the disaster database includes a plurality of key features affecting flood risk;

[0012] Screening module: used to screen the key features that affect flood risk, and construct a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socio-economic data and actual flood loss data;

[0013] Training module: used to construct multiple evaluation models and train them through the multidimensional feature data set, perform performance evaluation and explanatory analysis on the multiple evaluation models after training, and obtain evaluation results;

[0014] Evaluation module: used for selecting a final evaluation model based on the evaluation results, and performing flood loss risk evaluation through the final evaluation model.

[0015] In a third aspect, the present invention provides a data-driven flood insurance loss risk assessment device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, each step of the data-driven flood insurance loss risk assessment method as described in any one of the first aspects is implemented.

[0016] In a fourth aspect, the present invention further provides a storage medium having a computer program stored thereon, and when the computer program is executed, the various steps in the data-driven flood insurance loss risk assessment method as described in the first aspect are implemented.

[0017] The present invention provides a data-driven flood insurance loss risk assessment method and device, the method comprising: constructing a disaster database, the disaster database comprising a plurality of key features affecting flood risk; screening the key features affecting flood risk, and constructing a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socio-economic data and actual flood loss data; constructing a plurality of assessment models and training them through the multidimensional feature data set, performing performance evaluation and explanatory analysis on the trained plurality of assessment models to obtain assessment results; selecting a final assessment model based on the assessment results, and performing flood loss risk assessment through the final assessment model. Through the method provided by the present invention, by comprehensively analyzing meteorological, geographical, socio-economic data and actual flood loss data, a machine learning algorithm is used to efficiently evaluate flood risks. The model algorithm effectively handles incomplete or unstructured data, improves the flexibility and adaptability of the evaluation, and the interpretability analysis of the model increases decision makers' trust in the prediction results, promotes the effectiveness of the model in practical applications, thereby improving the generalization ability and prediction accuracy of the model, and helps to evaluate and manage flood risks more scientifically and accurately, and can provide a solid natural disaster science foundation for the stable operation and sustainable development of insurance companies. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. 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 the structures shown in these drawings without paying creative work.

[0019] Figure 1 It is a flow chart of the data-driven flood insurance loss risk assessment method of the present invention;

[0020] Figure 2 It is a schematic diagram of the program modules of the data-driven flood insurance loss risk assessment system of the present invention. DETAILED DESCRIPTION

[0021] In order to make the purpose, features and advantages of the invention more obvious and easy to understand, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0022] Please refer to Figure 1 , Figure 1 : is a flow chart of a data-driven flood insurance loss risk assessment method according to an embodiment of the present application. In this embodiment, the data-driven flood insurance loss risk assessment method includes:

[0023] Step 101: construct a disaster database, wherein the disaster database includes a plurality of key features affecting flood risks.

[0024] In this embodiment, the disaster database is used to store and manage various types of data related to flood disasters, specifically including multiple key features that affect flood risks.

[0025] Specifically, the disaster database is developed based on the EM-DAT (Emergency Events Database) database. Historical flood event data is collected from the EM-DAT database, and the data is cleaned and integrated to obtain the disaster database.

[0026] The EM-DAT database may also be replaced with customized commercial data to conduct respective insurance pricing and risk management analysis, thereby making the embodiment of the present application highly customizable and adaptable to different data characteristics and requirements, thereby providing a flexible risk assessment tool for insurance or reinsurance companies.

[0027] Step 102: Filter the key features that affect flood risk, and construct a multidimensional feature data set through the filtered key features, wherein the filtered key features at least include meteorological data, geographic information, socio-economic data, and actual flood loss data.

[0028] In this embodiment, key features that affect flood risks are screened to obtain screened key features, and then a multidimensional feature data set is constructed through the screened key features, wherein the screened key features include at least meteorological data, geographic information, socio-economic data and actual flood loss data, and the multidimensional feature data set is also constructed through the above-mentioned meteorological data, geographic information, socio-economic data and actual flood loss data.

[0029] Specifically, the key features that affect flood risk are quantitatively analyzed and selected through mutual information and variance inflation factor, so as to ensure that the multidimensional feature data set constructed by the selected key features can perform better in stability and predictive ability when training the evaluation model. Among them, mutual information (MI) is an important concept in information theory, which is used to measure the correlation between two random variables. The larger the mutual information, the stronger the correlation between the two random variables; the variance inflation factor (VIF) is a method used to detect whether there is a collinearity problem between independent variables in a multivariate regression model. The larger the VIF, the higher the degree of collinearity between independent variables.

[0030] Step 103: construct multiple evaluation models and train them using the multidimensional feature data set, perform performance evaluation and explanatory analysis on the trained multiple evaluation models, and obtain evaluation results.

[0031] In this embodiment, multiple evaluation models are first constructed, and the constructed multiple evaluation models are trained using the multidimensional feature data set constructed in step 102. Then, performance evaluation and explanatory analysis are performed on the multiple evaluation models trained using the multidimensional feature data set, so that evaluation results of the multiple evaluation models can be obtained.

[0032] Specifically, the construction algorithms of the multiple evaluation models include at least an integrated algorithm, a neural network model, a linear model and an example-based model, wherein the integrated algorithm includes random forest, gradient boosting tree, extreme gradient boosting, adaptive boosting, extreme random tree; the neural network model includes a multilayer perceptron; the linear model includes vector machine regression, Lasso regression, ridge regression; the example-based model includes K nearest neighbor. By constructing different types of evaluation models through the above different algorithms, the inherent patterns and associations of the data can be deeply explored, thereby enhancing the prediction ability of the model. Among them, the specific construction algorithm of the evaluation model is not limited here, and it is set according to actual needs.

[0033] Specifically, by implementing K-fold cross validation and grid search or random search methods on the multiple evaluation models after training, the performance of the multiple evaluation models after training is carefully evaluated and optimized, ensuring that the selected model has strong generalization ability and high prediction accuracy.

[0034] Specifically, by using SHAP and LIME tools to perform interpretative analysis on the multiple evaluation models after training, the prediction logic of the multiple evaluation models after training can be deeply analyzed, feature importance analysis from a global and local perspective is provided, transparency in the decision-making process of the evaluation model is provided, and the decision credibility of the end user is enhanced. Among them, SHAP (SHapley Additive Explanations) is one of the most popular XAI tools at present, which can explain various machine learning models, including tree models, linear models, deep learning models, etc. The core idea of ​​SHAP is to interpret the prediction results as the sum of the contribution values ​​of each feature. Specifically, SHAP can calculate the contribution value of each feature to the prediction result and present it in the form of SHAP value. The higher the SHAP value, the greater the influence of the feature on the prediction result. LIME (Local Interpretable Model-Agnostic Explanations) is an explanation method based on local linear models. LIME can explain any type of machine learning model, but its explanation effect is related to the complexity of the model. The working principle of LIME is to generate local perturbation data around the target prediction point, and use linear models to explain the impact of these perturbation data on the prediction results. The explanation focuses on the contribution of the variables to the outcomes, and continuous variables can also be split into categorical variables for explanation.

[0035] Step 104: Select a final assessment model based on the assessment results, and perform flood loss risk assessment using the final assessment model.

[0036] In this embodiment, based on the evaluation results obtained from the performance evaluation and explanatory analysis of multiple evaluation models, a final evaluation model is selected from multiple evaluation models for evaluating flood loss risks, wherein the final evaluation model selected according to the evaluation results has outstanding statistical performance and can provide clear feature explanations, thereby helping insurance or reinsurance industry practitioners to better understand and trust the prediction results of the model.

[0037] An embodiment of the present application provides a data-driven flood insurance loss risk assessment method, the method comprising: constructing a disaster database, the disaster database comprising a plurality of key features that affect flood risk; screening the key features that affect flood risk, and constructing a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socioeconomic data, and actual flood loss data; constructing a plurality of assessment models and training them through the multidimensional feature data set, performing performance evaluation and explanatory analysis on the trained plurality of assessment models to obtain assessment results; selecting a final assessment model based on the assessment results, and performing flood loss risk assessment through the final assessment model. Through the method provided by the present invention, by comprehensively analyzing meteorological, geographical, socio-economic data and actual flood loss data, a machine learning algorithm is used to efficiently evaluate flood risks. The model algorithm effectively handles incomplete or unstructured data, improves the flexibility and adaptability of the evaluation, and the interpretability analysis of the model increases decision makers' trust in the prediction results, promotes the effectiveness of the model in practical applications, thereby improving the generalization ability and prediction accuracy of the model, and helps to evaluate and manage flood risks more scientifically and accurately, and can provide a solid natural disaster science foundation for the stable operation and sustainable development of insurance companies.

[0038] Furthermore, the specific steps of the data-driven flood insurance loss risk assessment method in the embodiment of the present application are as follows:

[0039] 1. Collect historical flood event data from the EM-DAT database, clean and integrate the data, and then build a disaster database, which is used to store and manage various data related to flood disasters;

[0040] 2. Quantitatively analyze and select key features that potentially affect flood risk through mutual information and variance inflation factor, thereby constructing a multidimensional feature data set to ensure that the constructed model has excellent performance in stability and predictive ability;

[0041] 3. Select multiple machine learning algorithms to build multiple different evaluation models, and use multi-dimensional feature data sets to train the constructed multiple evaluation models;

[0042] 4. Use K-fold cross validation and network search or random search methods to carefully evaluate the performance of each evaluation model to ensure that the selected model has strong generalization ability and high prediction accuracy;

[0043] 5. Use SHAP and LIME tools to analyze the prediction logic of multiple evaluation models, provide feature importance analysis from a global and local perspective, provide transparency in the model decision process, and enhance the credibility of end-user decisions;

[0044] 6. Based on the performance evaluation and interpretative analysis results of multiple assessment models built with different algorithms in steps 4 and 5, select the most appropriate model algorithm for the final assessment of flood economic / insurance loss risk.

[0045] In summary, the embodiments of the present application provide a fast, accurate, and cost-effective method for assessing flood economic / insurance loss risks; through machine learning algorithms, incomplete or unstructured data is effectively processed, and the flexibility and adaptability of the assessment are improved; and the interpretability analysis of the model enhances the decision makers' trust in the prediction results and promotes the effectiveness of the model in practical applications; it also provides a new tool for the insurance / reinsurance industry, which helps to assess and manage flood risks more scientifically and accurately, and can provide a solid foundation of natural disaster science for the stable operation and sustainable development of insurance companies. At the same time, this method framework provides reinsurance companies with new tools for scientifically assessing risks and formulating corresponding reinsurance strategies, and provides a new perspective for flood risk managers.

[0046] Furthermore, the embodiment of the present application also provides a data-driven flood insurance loss risk assessment system 200. Figure 2 : is a schematic diagram of a program module of a data-driven flood insurance loss risk assessment system in an embodiment of the present application. In this embodiment, the data-driven flood insurance loss risk assessment system 200 includes:

[0047] Construction module 201: used to construct a disaster database, wherein the disaster database includes a plurality of key features affecting flood risk;

[0048] Screening module 202: used to screen the key features that affect flood risk, and construct a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socio-economic data, and actual flood loss data;

[0049] Training module 203: used to construct multiple evaluation models and train them through the multi-dimensional feature data set, and perform performance evaluation and explanatory analysis on the trained multiple evaluation models to obtain evaluation results;

[0050] Evaluation module 204: used to select a final evaluation model based on the evaluation results, and perform flood loss risk evaluation using the final evaluation model.

[0051] The embodiment of the present application provides a data-driven flood insurance loss risk assessment system 200, which can achieve: constructing a disaster database, the disaster database includes multiple key features that affect flood risk; screening the key features that affect flood risk, and constructing a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socio-economic data and actual flood loss data; constructing multiple assessment models and training them through the multidimensional feature data set, performing performance evaluation and explanatory analysis on the trained multiple assessment models to obtain assessment results; selecting a final assessment model based on the assessment results, and performing flood loss risk assessment through the final assessment model. Through the method provided by the present invention, by comprehensively analyzing meteorological, geographical, socio-economic data and actual flood loss data, a machine learning algorithm is used to efficiently evaluate flood risks. The model algorithm effectively handles incomplete or unstructured data, improves the flexibility and adaptability of the evaluation, and the interpretability analysis of the model increases decision makers' trust in the prediction results, promotes the effectiveness of the model in practical applications, thereby improving the generalization ability and prediction accuracy of the model, and helps to evaluate and manage flood risks more scientifically and accurately, and can provide a solid natural disaster science foundation for the stable operation and sustainable development of insurance companies.

[0052] Furthermore, the present application also provides a data-driven flood insurance loss risk assessment device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the various steps in the data-driven flood insurance loss risk assessment method described above are implemented.

[0053] Furthermore, the present application also provides a storage medium on which a computer program is stored. When the computer is executed by a processor, the various steps in the above-mentioned data-driven flood insurance loss risk assessment method are implemented.

[0054] Each functional module in each embodiment of the present invention may be integrated into a processing module, or each module may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may be stored in a computer-readable storage medium.

[0055] Based on this understanding, the technical solution of the specification of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., various media that can store program codes.

[0056] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present invention is not limited by the described order of actions, because according to the present invention, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention. In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0057] For those skilled in the art, according to the concept of the embodiments of the present application, there may be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A data-driven flood insurance loss risk assessment method, characterized in that: The method comprises: Building a disaster database, wherein the disaster database includes a plurality of key features affecting flood risk; Screening the key features that affect flood risk, and constructing a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socio-economic data, and actual flood loss data; Constructing multiple evaluation models and training them through the multidimensional feature data set, performing performance evaluation and explanatory analysis on the multiple evaluation models after training to obtain evaluation results; A final assessment model is selected based on the assessment results, and flood loss risk assessment is performed using the final assessment model.

2. The method according to claim 1, characterized in that The screening includes the following key features that affect flood risk: The key features affecting flood risk are quantitatively analyzed and selected through mutual information and variance inflation factor.

3. The method according to claim 1, characterized in that The construction algorithms of the multiple evaluation models include at least: random forest, gradient boosting tree, extreme gradient boosting, adaptive enhancement, extreme random tree; multi-layer perceptron; vector machine regression, Lasso regression, ridge regression; K nearest neighbor.

4. The method according to claim 1, characterized in that The performance evaluation and explanatory analysis of the multiple evaluation models after training include: The performance of the multiple evaluation models after training is evaluated and optimized by implementing K-fold cross validation and grid search or random search method on the multiple evaluation models after training.

5. The method according to claim 1, characterized in that The performing performance evaluation and interpretative analysis on the multiple evaluation models after training also includes: The SHAP and LIME tools are used to perform interpretative analysis on the multiple evaluation models after training, so as to analyze the prediction logic of the multiple evaluation models after training.

6. The method according to claim 1, characterized in that The disaster database is constructed based on the EM-DAT database, including: The historical flood event data in the EM-DAT database is collected, and the collected historical flood event data is cleaned and integrated to construct the disaster database.

7. The method according to claim 6, characterized in that The disaster database is constructed based on the EM-DAT database, and further includes: The EM-DAT database was replaced with customized commercial data.

8. A data-driven flood insurance loss risk assessment system, characterized in that: The system comprises: A construction module: used for constructing a disaster database, wherein the disaster database includes a plurality of key features affecting flood risk; Screening module: used to screen the key features that affect flood risk, and construct a multidimensional feature data set through the screened key features, wherein the screened key features at least include meteorological data, geographic information, socio-economic data and actual flood loss data; Training module: used to construct multiple evaluation models and train them through the multidimensional feature data set, perform performance evaluation and explanatory analysis on the multiple evaluation models after training, and obtain evaluation results; Evaluation module: used for selecting a final evaluation model based on the evaluation results, and performing flood loss risk evaluation through the final evaluation model.

9. A data-driven flood insurance loss risk assessment device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, each step of the data-driven flood insurance loss risk assessment method according to any one of claims 1 to 7 is implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, each step of the data-initiated flood insurance loss risk assessment method as described in any one of claims 1 to 7 is implemented.

Images