Intelligent division method for police case high-occurrence area based on density clustering and k-means clustering

Abstract

The application discloses an intelligent division method for police case high-occurrence areas based on density clustering and k-means clustering, and the method comprises the following steps: obtaining police case data of a city; preprocessing original police case data to obtain data after unification and standardization; clustering the data after preprocessing by using a density algorithm to obtain all categories and corresponding data of each category; clustering the police case data of all categories by using a k-means algorithm; then clustering all police cases of each police case data by using the density algorithm to obtain all categories and corresponding data of each category; extracting boundary points of each category data to obtain corresponding boundary contours, that is, to obtain police case high-occurrence areas. The application can accurately identify police case high-occurrence areas, and can effectively balance the uneven distribution of police case high-occurrence areas in urban areas and suburban areas, thereby providing effective data support for subsequent police deployment, unified dispatching and command.

Description

Technical Field

[0001] This invention belongs to the field of intelligent analysis technology of public safety data, and in particular relates to an intelligent classification method for high-crime areas based on density clustering and k-means clustering. Background Technology

[0002] Public safety is crucial to people's quality of life and social harmony and stability, and is especially important in densely populated urban areas. Timely and effective identification of high-crime areas can significantly reduce response time and improve call-handling efficiency.

[0003] Utilizing big data and artificial intelligence technologies to intelligently delineate key areas for police incidents is effective. Existing solutions, such as the method, device, electronic equipment, and storage medium for identifying high-incidence areas of police incidents disclosed in CN112131381A, are based on density clustering. While this method can effectively delineate key areas when police incidents are evenly distributed, it can lead to overly dense key areas in urban areas and omissions in key areas in suburban areas when police incidents are unevenly distributed between urban and suburban areas. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies by providing an intelligent method for classifying high-crime areas based on density clustering and k-means clustering. This method involves building a dataset from historical crime data using multiple dimensions such as timeline, weather, and holidays. First, density clustering machine learning algorithms are used to cluster the data. Then, k-means algorithms are used to cluster the crime data volume for all categories. Next, density clustering is applied again to all crimes within each category, resulting in all categories and their corresponding data. Finally, a rolling ball algorithm is used to extract boundary points for each category, ultimately obtaining information such as the latitude and longitude of the center point, boundary contour points, area, and number of crimes for each high-crime area. This invention is highly targeted, intelligent, and efficient, and to some extent solves the problem of uneven crime distribution between urban and suburban areas, where key urban areas are overly dense while key suburban areas are overlooked, providing strong data support for scientific police response and prevention.

[0005] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0006] The intelligent segmentation method for high-crime areas based on density clustering and k-means clustering of the present invention includes the following steps:

[0007] S1. Obtain police incident data for a certain city;

[0008] S2. Preprocess the original police report data to obtain data with the same standard. This step specifically involves:

[0009] S21. Data cleaning: Remove invalid alarm data to obtain information including alarm events, accuracy, dimensions, and alarm order numbers.

[0010] S22. Based on the time series, divide the data according to multiple dimensions including date and hour, weather, and holidays;

[0011] S3. Cluster the preprocessed data using a density algorithm to obtain all categories and the data corresponding to each category. Specifically, use the dbscan density clustering algorithm to cluster the preprocessed alarm data according to the set neighborhood distance and the minimum number of samples in the neighborhood of the core point to obtain all initial density clustering categories and the alarm data corresponding to each category.

[0012] S4. Use the k-means algorithm to cluster the alarm data of all categories. Specifically, use the k-means clustering algorithm to cluster the alarm data of all initial density clustering categories according to the set number of categories, and obtain all k-means clustering categories and the alarm data corresponding to each category.

[0013] S5. Then, for all the alarm data of each type, the density algorithm is used to cluster them to obtain all categories and the data corresponding to each category. Specifically, for the alarm data of each k-mean clustering category, the dbscan density clustering algorithm is used to cluster them according to the set neighborhood distance and the minimum number of samples in the neighborhood of the core point to obtain all the density clustering categories and the alarm data corresponding to each category.

[0014] S6. Extract boundary points for each category of data to obtain the corresponding boundary contour, which is the area with frequent police incidents. Specifically, the rolling ball method is used to extract boundary points for each secondary density clustering category of data to obtain the corresponding boundary contour, which is the final area with frequent police incidents.

[0015] The present invention has the following advantages over the prior art:

[0016] (1) Compared with existing technologies, this invention realizes the establishment of a dataset for historical police incident data according to multi-dimensional information such as time axis, weather, and holidays; firstly, the density clustering machine learning algorithm is used to cluster the data; then, the k-means algorithm is used to cluster the amount of police incident data in all categories; then, the density algorithm is used again to cluster all the police incidents in each category to obtain all categories and the data corresponding to each category; finally, the rolling ball algorithm is used to extract the boundary points of each category of data, and finally, the latitude and longitude of the center point, boundary outline points, area, number of police incidents, and other information of each police incident high-incidence area are obtained.

[0017] (2) This invention is highly targeted, intelligent and efficient. To a certain extent, it solves the problem of uneven distribution of police incidents in urban and suburban areas, resulting in overly dense key areas in urban areas and omissions in key areas in suburban areas. It provides strong data support for scientific police dispatch and prevention.

[0018] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the intelligent segmentation method for high-crime areas based on density clustering and k-means clustering of the present invention.

[0021] Figure 2 This is a partial latitude and longitude coordinate diagram showing the preprocessed police incident data for a certain area of ​​a certain city on September 10, 2022, in a specific embodiment.

[0022] Figure 3 This is an information diagram of the initial density clustering results corresponding to a specific embodiment;

[0023] Figure 4 The latitude and longitude coordinates of the initial density clustering results are shown in the specific embodiment.

[0024] Figure 5 This is a table showing the k-means clustering results in a specific embodiment;

[0025] Figure 6 A parameter combination diagram used in a specific embodiment;

[0026] Figure 7 This is a diagram showing the latitude and longitude coordinates of the results of the second density clustering in a specific embodiment;

[0027] Figure 8 The latitude and longitude coordinates of the boundary points of the clustering results of key areas of the police situation are shown in a specific embodiment. Detailed Implementation

[0028] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] The experimental tests for this invention were conducted using two 8-core Intel(R) Xeon(R) CPU E5630 @ 2.53GHz, two GeForce RTX 3090 GPUs, and Python version 3.7.9.

[0030] The following is in accordance with Figure 1 The flowchart of this invention details the steps involved in the technical solution. The intelligent classification method for high-crime areas based on density clustering and k-means clustering of this invention includes the following steps:

[0031] S1. Obtain police incident data for a certain city. For example, obtain at least the alarm time, number, longitude, and latitude from the police incident database of a certain district in a certain city.

[0032] S2. Preprocess the original police report data to obtain data with the same standard. This step specifically involves:

[0033] S21. Data cleaning: Remove invalid alarm data to obtain information including alarm events, accuracy, dimensions, and alarm order numbers.

[0034] S22. Based on the time series, the data is divided into multiple dimensions, including date and hour, weather, and holidays; The preprocessed latitude and longitude coordinates of some police incident data for a certain area of ​​a certain city on September 10, 2022 are shown below. Figure 2 As shown;

[0035] S3. Cluster the preprocessed data using a density algorithm to obtain all categories and the corresponding data for each category. In this specific embodiment, the dbscan density clustering algorithm is used to cluster the preprocessed alarm data based on a set neighborhood distance of 600 meters and a minimum number of samples in the neighborhood of the core point of 5, to obtain all initial density clustering categories and the corresponding alarm data for each category, such as... Figure 3 As shown in the figure. The latitude and longitude coordinates of the initial density clustering results are displayed as follows. Figure 4 As shown;

[0036] S4. Cluster the alarm data for all categories using the k-means algorithm. Specifically, the k-means clustering algorithm is used to cluster the alarm data for all initial density clustering categories according to a set number of categories, obtaining all k-means clustering categories and the alarm data corresponding to each category. In this specific embodiment, the k-means clustering algorithm is used to cluster the alarm data for all initial density clustering categories according to a set number of categories of 4, obtaining all k-means clustering categories and the alarm data corresponding to each category, such as... Figure 5 As shown;

[0037] S5. Then, for each category of alarm data, a density-based clustering algorithm is used to cluster all alarms, obtaining all categories and the corresponding data for each category. Specifically, for each k-means cluster category, the dbscan density clustering algorithm is used to cluster the alarm data based on the set neighborhood distance and the minimum number of samples in the neighborhood of the core point, obtaining all secondary density clustering categories and the corresponding alarm data for each category. In this specific embodiment, for each k-means cluster category, the dbscan density clustering algorithm is used to cluster the alarm data based on the set neighborhood distance and the minimum number of samples in the neighborhood of the core point, obtaining all secondary density clustering categories and the corresponding alarm data for each category. The parameter combination used in the example is as follows: Figure 6 As shown, the results of the second density clustering are as follows: Figure 7 As shown;

[0038] S6. Extract boundary points for each category of data to obtain the corresponding boundary contour, which is the high-crime area. Specifically, the alpha shape method is used to extract boundary points for each secondary density clustering category of data to obtain the corresponding boundary contour, which is the final high-crime area. Finally, information such as the latitude and longitude of the center point, boundary contour points, area, and number of incidents for each high-crime area is obtained. The final high-crime area clustering result boundary points are displayed as follows. Figure 8 As shown.

[0039] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for intelligent segmentation of high-crime areas based on density clustering and k-means clustering, characterized in that, Includes the following steps: S1. Obtain police incident data for a certain city; S2. Preprocess the original police report data to obtain data with the same standard. This step specifically involves: S21. Data cleaning: Remove invalid alarm data to obtain information including alarm events, accuracy, dimensions, and alarm order numbers. S22. Based on the time series, divide the data according to multiple dimensions including date and hour, weather, and holidays; S3. Cluster the preprocessed data using a density algorithm to obtain all categories and the data corresponding to each category. Specifically, use the dbscan density clustering algorithm to cluster the preprocessed alarm data according to the set neighborhood distance and the minimum number of samples in the neighborhood of the core point to obtain all initial density clustering categories and the alarm data corresponding to each category. S4. Use the k-means algorithm to cluster the alarm data of all categories. Specifically, use the k-means clustering algorithm to cluster the alarm data of all initial density clustering categories according to the set number of categories, and obtain all k-means clustering categories and the alarm data corresponding to each category. S5. Then, for all the alarm data of each type, the density algorithm is used to cluster them to obtain all categories and the data corresponding to each category. Specifically, for the alarm data of each k-mean clustering category, the dbscan density clustering algorithm is used to cluster them according to the set neighborhood distance and the minimum number of samples in the neighborhood of the core point to obtain all the density clustering categories and the alarm data corresponding to each category. S6. Extract boundary points for each category of data to obtain the corresponding boundary contour, which is the area with frequent police incidents. Specifically, the rolling ball method is used to extract boundary points for each secondary density clustering category of data to obtain the corresponding boundary contour, which is the final area with frequent police incidents.

Images