A POI-associated building area regression prediction device and method

Through the POI-associated building area regression prediction method, the regression analysis model and random forest algorithm are used to consider multiple POI feature data, which solves the problem that POI point-associated building area in the existing technology and improves the accuracy of building area prediction , especially in education and culture, the prediction effect has been significantly improved.

CN113360966BActive Publication Date: 2024-11-12CHINA REINSURANCE (GROUP) CORPORATION +1
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202110390924.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-04-12
Publication Date
2024-11-12
Estimated Expiration
2041-04-12

AI Technical Summary

Technical Problem

The existing building area prediction work is difficult to accurately obtain the building area associated with POI points, which affects the construction of the risk exposure database and the accuracy of catastrophe model loss analysis.

Method used

The POI-associated building area regression prediction method is adopted, and the POI-associated building area is predicted by obtaining the POI data to be tested, using the regression analysis model to predict the associated building area, taking into account the characteristic data such as POI point type, importance, grading, geographical administrative division, night light index and population density, combined with Random forest algorithm makes predictions.

Benefits of technology

A good prediction of unknown building area has been achieved, especially in education and culture prediction results have been significantly improved, and the overall effect of building area prediction model has been improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure BDA0003016676830000041
    Figure BDA0003016676830000041
  • Figure BDA0003016676830000042
    Figure BDA0003016676830000042
  • Figure BDA0003016676830000043
    Figure BDA0003016676830000043
Patent Text Reader

Abstract

The present invention relates to a POI associated building area regression prediction method, comprising the following steps: S1. Acquire POI data to be tested; S2. Predict using a regression analysis model, input the POI data to be tested into a regression analysis model trained by an algorithm using POI feature data to predict associated building area data related to the POI data to be tested, wherein the POI feature data includes but is not limited to POI point type (kind), POI point importance (imprtnc), POI classification (POI_flg), number of subclass POI points (snClssN), geographical administrative division (CODE) where the POI point is located, night light index (NTL) corresponding to the POI point, population density (POP), and land use type (LU); S3. Obtain the predicted associated building area data. The overall effect of this building area prediction model is good, and it can be used to predict the associated building area of ​​the POI point to be predicted with an unknown building area value, especially for the education and culture category.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of catastrophe model loss analysis and calculation, and in particular relates to a POI-associated building area regression prediction device and method. Background Art

[0002] Part of the existing building area prediction work is aimed at predicting per capita housing area, for example [1] (Ye Chun, Wu Xianghua, Huang Yuying. Research on prediction of per capita housing building area in urban areas - taking Jiangsu Province as an example [J]. China Real Estate, 2020 (18): 73-79.). This work predicts per capita housing building area in urban areas. It selects three major influencing factors: economic and social development, changes in population and family size, and development of the housing market. The influencing factors are quantified, and a multivariate linear regression and gray system prediction model is constructed to predict the per capita housing building area in urban areas of Jiangsu Province in the next five years. This part of the work only involves the prediction of residential housing area, and does not include predictions of other types of building area. Another part of the work focuses on the prediction of housing area in real estate surveying and mapping [2] (Wang Jiakuan. Comparative analysis of housing area prediction and actual measurement in real estate surveying and mapping [J]. Science and Technology Innovation, 2017 (27): 186-187.). This type of work does not use machine learning models for regression analysis. It relies on the experience and skills of surveying and mapping personnel to accurately grasp the key points of housing surveying and mapping work, predict housing area, and avoid deviations in surveying and mapping results. Currently, there is no prediction target for the building area associated with POI points. Regression analysis models are used to serve the construction of risk exposure industry databases and assist in the work of reinsurance underwriting. Summary of the invention

[0003] In order to solve the technical problem proposed by the present invention, the present invention provides a POI-associated building area regression prediction method, comprising the following steps:

[0004] S1. Obtain POI test data;

[0005] S2. Using the regression analysis model to predict, the POI data to be tested is input into the regression analysis model trained by the algorithm based on the POI feature data to predict the associated building area data related to the POI data to be tested, wherein the POI feature data includes but is not limited to the POI point type (kind), POI point importance (imprtnc), POI classification (POI_flg), number of sub-class POI points (snClssN), geographical administrative division (CODE) where the POI point is located, night light index (NTL) corresponding to the POI point, population density (POP), and land use type (LU);

[0006] S3. Obtain the predicted associated building area data.

[0007] Another aspect of the present invention provides a POI-associated building area regression prediction device, the device comprising at least one processor; and

[0008] The memory stores instructions, and when the instructions are executed by at least one processor, the method provided by the present invention is implemented.

[0009] The beneficial effect of the present invention is that the building area prediction model has a good overall effect and can be used to predict the associated building areas of POI points to be predicted with unknown building area values, especially for education and culture. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 .Schematic diagram of the experimental process. DETAILED DESCRIPTION

[0011] The risk exposure industry database AED (Aggregate Exposure Database) is an important part of the catastrophe model loss analysis and calculation, which can be used to estimate the insurable loss (Insurable Loss) and insured loss (Insured Loss) of residential, industrial and commercial buildings caused by disasters. When the direct insurance company can only provide a regional total insurance amount, AED can assist the catastrophe model to distribute the insurance amount as accurately as possible in different spatial locations, thereby providing an auxiliary reference for reinsurance underwriting. Point of Information (POI), in the geographic information system, can be a house, a shop, a mailbox, a bus stop, etc. In terms of spatial distribution, each building has a POI point closest to it. Such a building is called a POI associated building, and each POI point may correspond to multiple associated buildings. The associated building area of ​​POI points is an important indicator for the construction of risk exposure databases. However, it is difficult to accurately obtain the associated building area (referred to as building area) of each POI point through existing equipment and resources. This becomes a thorny problem in the process of building a risk exposure database. The solution provided by the present invention is described below in conjunction with specific embodiments, but should not be limited by specific embodiments.

[0012] In some embodiments, the POI-associated building area regression prediction method includes the following steps:

[0013] S1. Obtain POI test data;

[0014] S2. Using the regression analysis model to predict, the POI data to be tested is input into the regression analysis model trained by the algorithm based on the POI feature data to predict the associated building area data related to the POI data to be tested, wherein the POI feature data includes but is not limited to the POI point type (kind), POI point importance (imprtnc), POI classification (POI_flg), number of sub-class POI points (snClssN), geographical administrative division (CODE) where the POI point is located, night light index (NTL) corresponding to the POI point, population density (POP), and land use type (LU);

[0015] S3. Obtain the predicted associated building area data.

[0016] In some embodiments, the output of step S2 is the associated building area weight. The building area value of the POI point can be obtained by multiplying the total building area value of the area with a known total building area by the area ratio of the POI point in the area.

[0017] In some embodiments, the building area associated with the POI points predicted by the prediction set is used as the building area weight. In other embodiments, the total building area of ​​the corresponding category that is easily available can be multiplied by the predicted building area ratio of the corresponding POI point (calculated according to formula 1) to obtain a more accurate building area value. The C1 category can be any of the eight categories of wholesale and retail, education and culture, health and social security, sports and leisure, public facilities, companies and enterprises, transportation and warehousing, and scientific research and technology.

[0018]

[0019] Combination Figure 1 As shown, the following is a specific implementation example of the training model.

[0020] 1. The original data is shown in Table 1:

[0021] Table 1. Raw data

[0022]

[0023] 2. Data field description:

[0024]

[0025]

[0026] 3.1 LU land use types include:

[0027]

[0028] The training data of the 68 core cities includes all other categories except the 401 category.

[0029] 3.2 POI type, kind field: including 16 categories such as catering, accommodation, wholesale and retail, automobile sales and services, finance and insurance, education and culture, health and social security, sports and leisure, public facilities, commercial facilities and business services, resident services, companies and enterprises, transportation and warehousing, scientific research and technical services, agriculture, forestry, animal husbandry and fishery, and natural features\name. The training data of 68 core cities this time includes 8 categories such as wholesale and retail, education and culture, health and social security, sports and leisure, public facilities, companies and enterprises, transportation and warehousing, and scientific research and technology.

[0030] 3.3 POI classification, Poi_flg field includes:

[0031]

[0032] The training data of 68 core cities include categories 1, 2, 3, 4, 5, and 6.

[0033] 3.4 List of core cities:

[0034]

[0035] 4. Data preprocessing methods and feature design:

[0036] 4.1 Add the first 3 fields of kind to identify major categories such as catering, accommodation, wholesale and retail, education and culture

[0037] 4.2 Add the first digit of the code to identify the region

[0038] 4.3 Add the first two digits of the code to indicate the province

[0039] 4.4 Add the first 4 characters of the code to identify the city

[0040] 4.5 NTL value rounding

[0041] 4.6 POP is divided into segments by dividing by 50

[0042] In some other embodiments, other feature processing methods are used (changing int(NTL), int(pop / 50) to other processing methods), but the experimental results are not as good as the current feature processing method.

[0043] 5. The input data form after preprocessing:

[0044]

[0045] Feature combination: kind, the first 3 digits of kind, imprtnc, poi_flg, snCl ssN, int(NTL), int(pop / 50), LU, code, the first digit of code, the first 2 digits of code, the first 4 digits of code.

[0046] 6. Regression analysis using random forest algorithm

[0047] The experimental effect is evaluated by 5-fold cross validation on the training set. The specific process is that the random forest algorithm directly calls the random forest regression analysis model in the weka platform, see the configuration (100, 100, 100, 1, 2, 0, 1) in Table 1. Surprisingly, this set of parameters can achieve the best experimental effect among the several parameter configurations tried. That is, in Table 1, the original parameter configuration is (100, 100, 100, 1, 2, 0, 1), and the experimental results in Tables 6 and 7 in the following "7 Experimental Results" can be obtained. However, when parameter configuration 1 is adjusted to (100, 100, 90, 1, 2, 0, 1), the experimental results are shown in Table 2; when parameter configuration 2 is adjusted to (90, 100, 90, 1, 2, 0, 1), the experimental results are shown in Table 3; when parameter configuration 3 is adjusted to (100, 100, 110, 1, 2, 0, 1), the experimental results are shown in Table 4. The experimental results obtained by trying configuration 1, configuration 2 and configuration 3 are all worse than the experimental results in “7 Experimental Results”. It can be seen that the optimal parameter configurations in Table 6 and Table 7 are not obvious.

[0048] Table 1 Random forest algorithm model parameters (parameter configuration)

[0049] parameter value Parameter Description bagSizePercent 100 The size of the bag, expressed as a percentage of the training set batchSize 100 The number of samples selected for one training numIterations 100 The number of trees in the random forest seed 1 Random seed trees used numDecimalPlaces 2 The number of decimal places used for numbers in the model maxDepth 0 Maximum tree depth, set to 0 for unlimited numExecutionSlots 1 Number of threads

[0050] Table 2 Parameter configuration 1 Experimental results

[0051]

[0052] Table 3 Experimental results of parameter configuration 2

[0053]

[0054] Table 4 Experimental results of parameter configuration 3

[0055]

[0056] 7 Experimental results:

[0057] Table 6 Test case parameter configuration experimental results

[0058]

[0059] Table 7 Test case parameter configuration experimental results

[0060]

[0061] Experimental results analysis: There are 44,826 data from 68 core cities across the country, including 8 categories such as wholesale and retail, education and culture, health and social security, sports and leisure, public facilities, companies and enterprises, transportation and warehousing, scientific research and technology, and building area prediction, with a correlation coefficient of 0.56798. The model effect is good. Education and culture data, including 19,158 records, has the largest amount of data in a single category, and the experimental effect is greatly improved, with a correlation coefficient of 0.695923. This shows that expanding the data scale can improve the experimental effect.

[0062] The test case uses the random forest algorithm combined with different features to compare the test results. The 5-fold cross validation method is used to evaluate the experimental results.

[0063] In the comparison table of random forest regression analysis results based on different features, all features are used to give the regression results first, and then feature subtraction is performed to subtract a single feature or feature group. Surprisingly, "-LC_aggr" (ratio of impermeable layer (%)) with ID 2 is the optimal feature configuration, which has achieved unexpected technical results, with a correlation coefficient of 0.56798, see Table 8.

[0064] Table 8 Comparison of random forest regression analysis results based on different features

[0065]

[0066]

[0067] The embodiments and functional operations of the subject matter described in this specification may be implemented in: digital electronic circuitry, tangibly implemented computer software or firmware, computer hardware, including the structures disclosed in this specification and their structural equivalents, or a combination of more than one of the above. The embodiments of the subject matter described in this specification may be implemented as one or more computer programs, i.e., one or more modules of computer program instructions encoded on one or more tangible non-transitory program carriers, for execution by a data processing device or to control the operation of a data processing device.

[0068] A computer program (which may also be referred to or described as a program, software, software application, module, software module, script or code) may be written in any form of programming language, including compiled or interpreted languages ​​or declarative or procedural languages, and the computer program may be deployed in any form, including as a stand-alone program or as a module, component, subroutine or other unit suitable for use in a computing environment. A computer program may, but need not, correspond to a file in a file system. A program may be stored in a portion of a file that stores other programs or data, for example, one or more scripts stored in the following: in a markup language document; in a single file dedicated to a related program; or in multiple collaborative files, for example, files storing one or more modules, subroutines or code portions. A computer program may be deployed to execute on one or more computers, the computers being located in one place or distributed to multiple locations and interconnected by a communication network.

[0069] The processes and logic flows described in this specification may be performed by one or more programmable computers that execute one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows may also be performed by special purpose logic circuits, such as FPGAs (field programmable gate arrays) or ASICs (application specific integrated circuits), and the apparatus may also be implemented as special purpose logic circuits.

[0070] The computer suitable for implementing the computer program includes and exemplarily can be based on a general-purpose microprocessor or a special-purpose microprocessor or both of the above-mentioned processors, or any other kind of central processing unit. Generally, the central processing unit will receive instructions and data from a read-only memory or a random access memory or both. The main element of the computer is a central processing unit for running or executing instructions and one or more memory devices for storing instructions and data. Generally, the computer will also include or be operably coupled to receive data from one or more large-capacity storage devices for storing data or transfer data to the large-capacity storage device, or receive and transfer both, and the large-capacity storage device is, for example, a magnetic disk, a magneto-optical disk or an optical disk. However, the computer does not necessarily have such a device. In addition, the computer can be embedded in another device, for example, a mobile phone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a global positioning system (GPS) receiver or a removable storage device, for example, a universal serial bus (USB) flash drive, etc.

[0071] Computer readable media suitable for storing computer program instructions and data include all forms of nonvolatile memory, media and memory devices, including, as examples: semiconductor memory devices, such as EPROM, EEPROM and flash memory devices; magnetic disks, such as internal hard disks or removable disks; magneto-optical disks; CD-ROM and DVD-ROM disks. The processor and memory may be supplemented by or incorporated into a special purpose logic circuit.

[0072] Although this specification contains many specific implementation details, these should not be interpreted as limitations on the scope of any invention or the scope of what can be claimed, but rather as an illustration of features that can concretize a particular embodiment of a particular invention. The specific features described in this specification in the context of independent embodiments can also be implemented in combination with a single embodiment. Conversely, the various features described in the context of a single embodiment can also be implemented independently in multiple embodiments, or in any suitable sub-combination. In addition, although the features can be described above as a combination effect and even initially required, one or more features from the required combination can be removed from the combination in some cases, and the required combination can be turned to a sub-combination or a variation of the sub-combination.

Claims

1. A POI-associated building area regression prediction method, characterized in that: The steps include: S1. Obtain POI test data; S2. Using the regression analysis model to predict, the POI data to be tested is input into the regression analysis model trained by the algorithm based on the POI feature data to predict the associated building area data related to the POI data to be tested, wherein the POI feature data includes POI point type, POI point importance, POI classification, number of sub-class POI points, geographical administrative division where the POI point is located, night light index corresponding to the POI point, population density, and land use type; S3. Obtaining the predicted associated building area data; The output of step S2 is the associated building area weight. The building area value of the POI point can be obtained by multiplying the total building area value of the area with a known total building area by the area ratio of the POI point in the area. The algorithm of step S2 includes a random forest algorithm.

2. The method according to claim 1, characterized in that The random forest algorithm is a random forest regression analysis model in the weka platform that is directly called, and the parameter values ​​are: bagSizePercent 100, batchSize 100, numIterations 100, seed1, numDecimalPlaces 2, maxDepth 0, and numExecutionSlots 1.

3. The method according to claim 2, characterized in that The training method of the regression analysis model in step S2 includes data preprocessing and feature design steps: Add the first 3 digits of the POI point type to identify the major categories, including catering, accommodation, wholesale and retail, and education and culture; Add the first position of the geographical administrative division where the POI point is located to identify the region; Add the first two digits of the geographical administrative division where the POI point is located to identify the province; Add the first four digits of the geographical administrative division where the POI point is located to identify the city; The night light index value corresponding to the POI point is rounded; Population density is divided into segments by dividing by 50.

4. The method according to claim 3, characterized in that The input data form and feature combination method after the data preprocessing include: POI point type, the first 3 POI point types, POI point importance, POI classification, number of sub-category POI points, int (night light index corresponding to the POI point), int (population density / 50), land use type, the geographical administrative division where the POI point is located, the first geographical administrative division where the POI point is located, the first 2 geographical administrative divisions where the POI point is located, and the first 4 geographical administrative divisions where the POI point is located.

5. A POI-associated building area regression prediction system, characterized in that: The system comprises at least one processor; and a memory storing instructions, which, when executed by the at least one processor, implement the steps of the method according to any one of claims 1 to 4.

6. A computer program product comprising a computer program / instructions, characterized in that When the program is executed by at least one processor, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by at least one processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Citation Information

Patent Citations

  • Method and system for estimating population density in urban built-up areas by cooperation with spatial big data

    CN108871286A

  • Remote sensing image urban building function classification method

    CN110147779A

  • Remote sensing image building area land utilization attribute space migration method

    CN110472559A

  • Population spatial distribution simulation method and device based on urban inland inundation risk assessment

    CN111881573A

  • Population data spatialization method, device and equipment based on multi-source information fusion

    CN112395383A