JP7292838B2 - Incident occurrence deterrence effect prediction system - Google Patents

Description

The present invention relates to an incident prevention effect prediction system.

A method has been proposed for predicting the location and time of crime by applying a template matching method to personal position statistical information created based on each person's position information and attribute information (Patent Document 1).

A crime prediction system has also been proposed that determines crime predictions by adjusting past crime rates based on correlations between weather conditions and crime data (see US Pat.

JP 2016-166938 (Patent 5992563) Japanese Patent Publication No. 2018-505474

However, when it is assumed that the occurrence of incidents can be prevented by proactive measures such as the police's vigilance activities, it is necessary to consider the deterrent effect on the occurrence of incidents caused by the police's warning. If a police officer's vigilance site is determined without estimating the police's vigilance effect, there is a possibility of overlooking a site that is highly effective in deterring the occurrence of an incident.

Patent Literatures 1 and 2 do not take into consideration the effect of deterring the occurrence of an incident caused by a police officer's warning.

SUMMARY OF THE INVENTION An object of the present invention is to effectively determine the destination of a warning by predicting the effect of deterring the occurrence of an incident caused by a police officer's warning.

An incident occurrence deterrence effect prediction system according to one aspect of the present invention includes a data server having a database storing data used for predicting incident occurrence, and learning the data stored in the database to predict the incident occurrence. A learning server for constructing a model, a prediction server for predicting the deterrence effect amount of the occurrence of the proposal by referring to the prediction model, and prediction by the prediction server by referring to the input data entered by the user and a user terminal for visualizing and displaying the predicted effect amount obtained by the training on a map. constructing the prediction model for generation of the proposal, and the prediction server suppresses the generation of the proposal using the prediction model constructed by the learning server and the input data input from the user terminal; The amount of effect is obtained, and the user terminal visualizes and displays the amount of deterrence effect of generating the proposal on the map.

According to one aspect of the present invention, it is possible to effectively determine the destination of the warning by predicting the effect of deterring the occurrence of an incident caused by the warning by the police officer.

1 is a schematic diagram of the overall configuration of an incident occurrence deterrence effect prediction system of Embodiment 1; FIG. (a) is a figure which shows the functional structure of the server for learning, (b) is a figure which shows the functional structure of the server for prediction. FIG. 10 is a time-series sequence diagram showing exchanges between servers of the incident prevention effect prediction system; FIG. 4 is a diagram schematically showing an example of a learning input table format used for constructing an incident prediction model; 10 is a flowchart showing learning processing for creating a prediction model that outputs incident occurrence probabilities. FIG. 10 is a diagram schematically showing an example of a prediction input table format used to calculate incident occurrence probabilities for each of the presence or absence of the implementation of the police, (a) is a schematic diagram in the case of implementing the police, (b) ) is a schematic diagram when the guard is not implemented. It is a flowchart which shows the prediction process for predicting the incident occurrence deterrence effect amount by the police. It is a figure which shows the graphical user interface of an incident occurrence deterrence effect prediction system. FIG. 12 is a diagram schematically showing an example of a case priority weight table defining priority case types according to the second embodiment; FIG. 11 is a functional configuration diagram of a prediction server 120 to which a case priority weight storage unit 1000 of Example 2 is added; 10 is a flow chart for calculating alert priorities for determining alert destinations in Example 2. FIG. FIG. 11 is a diagram schematically showing an example of a learning input table format used for constructing an incident occurrence prediction model including a warning type variable in Example 3; 13 is a flow chart showing prediction processing for predicting an incident occurrence deterrence effect amount for each type of warning according to the third embodiment. FIG. 2 is an overall flow schematic diagram schematically showing the overall flow of learning and prediction; FIG. 11 is a diagram schematically showing an example of a table defining embodiment patterns of the police of Example 3;

An embodiment will be described below with reference to the drawings.

With reference to FIG. 1, the overall configuration of the incident prevention effect prediction system of the first embodiment will be described.
The incident deterrence effect prediction system of the first embodiment basically comprises a data server 100 having various databases used for prediction, a learning server 110 for learning data and constructing a prediction model, and predicting an incident deterrence effect. and a computer 130 (user terminal) having an application for visualizing prediction results on a map.

The main databases used in the incident deterrence effect prediction system are an incident database 101 that stores information such as the date and time of incident occurrence, location, type of incident, etc., and a geographic information database 102 that stores information such as population statistics and facilities. Also, there is a GPS database 103 storing position information of police officers and a map database 104 storing map information for visualization. However, in addition to these, for example, there may be a demographic database based on location information of mobile phones and the like.

The case database 101 stores at least information such as a case ID, the date, time, location, and type of case when the case occurred. Case types include, for example, traffic accidents, murders, burglaries, and bicycle thefts.

The geographic information database 102 contains demographic data including age-specific population ratios, household ratios, etc., locations and usage of facilities (for example, police stations, police stations, stations, schools, etc.), land use categories (for example, residential areas). , fields, roads, forests, etc.). Information such as weather, temperature, amount of rainfall, occurrence of sporting events, etc. may also be included as long as it can correspond to geographical location information.

The GPS database 103 stores location information acquired by a location measurement function provided in a terminal possessed by a police officer and a device mounted in a vehicle. This location information includes at least timestamp, latitude, and longitude information. In addition to these three elements, it may also include information such as the type of work of the police officer and the type of vehicle.

The map database 104 is map data for a geographical information system (GIS), which is used to display the predicted location of an incident on the application. It should be noted that the map data does not necessarily have to be possessed in the form of the map database 104 as long as it can be obtained separately online.

The learning server 110 and the prediction server 120 receive necessary data from the data server 100, process the data into an input table format, and then perform calculation processing by machine learning. The learning server 110 learns geographic information and police position information data using incident occurrence as teacher data, and constructs an incident occurrence prediction model. The prediction server 120 uses the incident occurrence prediction model constructed by the learning server 110 and the user input data sent from the computer 130 used by the user to finally output the prediction result of the incident occurrence deterrence effect. This allows the user to confirm from the computer 130 the locations that are expected to be highly effective in deterring the occurrence of incidents.

A functional configuration (a) of the learning server 110 and a functional configuration (b) of the prediction server 120 will be described with reference to FIG. 2 .
As shown in FIG. 2(a), the learning server 110 has an input unit 200 for inputting data obtained from the data server 100, and processes the input data into an input table format as shown in FIG. A data processing unit 201, an input table storage unit 205 that stores an input table, a learning control unit 202 that controls the timing of executing learning, a prediction model construction unit 203 that executes learning, and outputs the constructed prediction model. and a prediction model storage unit 206 that stores the prediction model.

In order to predict the effect of deterring the occurrence of incidents for each time and place, the data processing unit 201 aggregates various data for each prediction unit. As for the location, the prediction target geographic range is meshed, and an ID is assigned to each mesh. The mesh is divided into, for example, 100 m square meshes, but the mesh size may be larger or smaller. Predict locations with high/low deterrence effect for incident occurrence for each set mesh. Similarly, the time is also divided into a certain time zone size and predicted. For example, prediction every two hours is taken as an example, but the width of two hours is just an example, and it may be longer or shorter than this.

Next, as shown in FIG. 2B, the prediction server 120 includes an input unit 210 that receives data acquired from the data server 100 and a prediction model acquired from the learning server 110, and a prediction request from the user. and a request reception unit 215 that receives input parameters, a data processing unit 211 that processes data and input parameters into an input table format as shown in FIG. 6, a prediction control unit 212 that controls prediction execution for each prediction condition, A prediction processing unit 213 that performs deterrent effect prediction processing using the prediction model and the input table received from the learning server 110, an output unit 214 that outputs prediction results, and a prediction result storage unit 216 that stores the prediction results. , provided.

It is not always necessary to execute the prediction process every time a prediction request is received from the user, and for already predicted conditions, the prediction results stored in the prediction result storage unit 216 may be output.

Next, with reference to the time-series sequence of FIG. 3, the flow of communication between each server will be described.
As shown in FIG. 3, the learning server 110 receives data as input from the data server 100, performs learning processing, and outputs a prediction model. The prediction server 120 receives as inputs a prediction model from the learning server 110, data from the data server 100, and input parameters from the computer 130 operated by the user. These inputs are used to calculate the predicted value of the deterrent effect, and the predicted result is output to the computer 130 side.

Since learning is a time-consuming process, basically, once a prediction model is constructed, prediction processing is executed using the same prediction model in response to a prediction request from a user. However, the learning control unit 202 may perform re-learning when a decrease in prediction accuracy is confirmed or at regular timings.

Based on the above premise, a method for predicting the effect of deterring the occurrence of incidents will be described with reference to the overall flow of FIG. 14 .
As shown in FIG. 14, first, in learning, various data are processed to create a learning input table, and machine learning is used to learn the relationship between past geographic information, police implementation status, and incident occurrence. A prediction model 1400 is constructed and output by this learning.

In the prediction, at future dates and times, the probability of incident occurrence for each mesh is predicted when vigilance is performed and when vigilance is not performed. These are output by creating two types of prediction input tables in which only variable values are changed and inputting them to the prediction model 1400 described earlier. Finally, the difference in the incident occurrence probability due to the presence or absence of the police is regarded as the deterrent effect of the police officer's warning, and this is output as the deterrent effect amount.
Individual details of the above overall flow will be described below.

FIG. 4 is a diagram showing an example of the format of a learning input table used during learning.
As shown in FIG. 4, the learning input table holds mesh IDs, dates, time zones, geographic information variable groups, police variables, and event variables as columns. Each geographic information variable is information associated with a mesh ID, or a mesh ID and a date, or a mesh ID and a time zone, and is acquired from the geographic information database 102 . The event variable is teacher data having a value of 1 when an event occurs and 0 when an event does not occur for each mesh ID, date, and time zone. Whether or not an incident has occurred is aggregated from information on the date and time and place of occurrence of each incident in the incident database 101 .

The alert variable is a variable having a value of 1 if the alert was performed before the occurrence of the incident and a value of 0 if the alert was not performed for each mesh ID, date, and time period. The presence or absence of police presence is aggregated from position information and time stamp information of police officers or vehicles in the GPS database 103 . Note that mesh IDs and dates are not used as feature amounts in machine learning.

A process of constructing a prediction model 1400 by machine learning in the learning server 110 will be described with reference to the flowchart of FIG. 5 during learning. In this learning, the presence or absence of incidents is used as teacher data, and a model is constructed that predicts the occurrence of future incidents based on geographic information and information on police operations.

First, the input unit 200 acquires case, geography, and GPS information from each database (S500). When the data is acquired, the data processing unit 201 processes the data into the format of the learning input table shown in FIG. 4 (S501).

Next, when the preparation of the input table for learning is completed, the prediction model construction unit 203 constructs a prediction model by machine learning using the case variable as teacher data, and the time period, geographic information, and police variables as feature vectors. (S502). Examples of machine learning algorithms here include logistic regression models. However, as long as the algorithm can express the output as a probability value, it is not limited to logistic regression. By this processing, an incident occurrence prediction model is generated that outputs a probability value of being classified into an incident occurrence class when given a feature amount of a certain mesh in a certain time period in the future.

Finally, the output unit 204 saves the incident occurrence prediction model and outputs it to the prediction server 120 (S503).

Next, a method for predicting the effect of deterring the occurrence of incidents in the prediction server 120 will be described. FIG. 6 is a diagram showing an example of the format of a prediction input table used for prediction.
Similar to the learning input table (see FIG. 4), the prediction input table holds mesh IDs, dates, time zones, geographic information variable groups, and warning variables as columns. Since event variables have unknown values at the time of prediction, they do not exist in the prediction input table. Two types of prediction input tables having the above format are prepared, one for when the warning is executed (a) and the other when the warning is not executed (b). The mesh ID, the date, and the time zone are the values of the place and the date and time to be predicted, respectively, and are set to the same values in the two prediction input tables. Since each geographic information variable is also a value corresponding to a place or date and time, the two prediction input tables have the same value.

What differs between the two prediction input tables is the value of the warning variable. All 1's are stored in the warning variables of the prediction input table (a), which is the case where the warning is executed for the place and the date and time to be predicted. On the other hand, all 0s are stored in the warning variables of the prediction input table (b), which is the case where the warning is not executed. Using the two prediction input tables created here, the deterrent effect of the police is predicted.

A process of predicting the effect of deterring the occurrence of incidents by machine learning executed by the prediction server 120 will be described with reference to the flowchart of FIG.
First, the input unit 210 acquires a prediction model of incident occurrence from the learning server 110 (S700). Also, the request receiving unit 215 acquires input parameters from the computer 130 terminal operated by the user (S700). The input parameters include the date and time period for which the user wants to make a prediction, the forecast weather information for that time period, the type of incident to be predicted, and the like. However, the user's input may be omitted by automatically obtaining information such as the weather online. Then, the input unit 210 again acquires the case, geography, and GPS information from each database (S700).

Next, the data processing unit 211 creates two types of prediction input tables shown in FIG. 6 (S701). After creating the prediction input table, the prediction processing unit 213 inputs the two types of prediction input tables to the prediction model, respectively, and calculates the incident occurrence probability when vigilance is performed in each mesh and the incident occurrence probability when vigilance is not performed. Each probability is calculated (S702). Then, the difference in incident occurrence probability depending on whether or not the police are implemented is calculated, and this is defined as the amount of deterrent effect by the police (S702). The amount of deterrence effect may be defined not only as a difference but also as a reduction rate of the incident occurrence probability from the case where the vigilance is not implemented to the case where the vigilance is implemented.

Finally, the output unit 214 saves the amount of deterrent effect and outputs it to the computer 130 terminal used by the user (S703).

FIG. 8 is an example of a user interface on an application on the computer 130 operated by the user.
As shown in FIG. 8, a map 800 obtained from the map database 104 is displayed on the screen. The user selects a date selection button 801 for selecting the date for which the deterrent effect of incident occurrence is to be predicted, a time period selection button 802 for selecting the time period for which the prediction is desired, a weather selection button 803 for selecting forecast weather information for the desired date and time, A case selection button 804 for selecting the type of case to be predicted is pressed to set, and a prediction display button 805 is pressed. Then, the prediction result of the deterrent effect amount for each mesh is output from the prediction server 120 , and the prediction result 806 for each mesh is visualized on the map 800 by expressing it in shades of color.

In addition, the map can be enlarged or reduced for display using a scale setting bar 807, and the mesh size is enlarged or reduced according to the enlargement or reduction of the map.

Since the prediction result 806 for each mesh is determined based on the predicted value of the deterrent effect amount, the color gradation may be displayed in a plurality of stages. Alternatively, the predicted values may be displayed directly on each mesh. Alternatively, the predicted value may be balloon-displayed only when the mouse cursor is placed on the mesh.

The mesh size may be dynamically changed according to the enlargement and reduction of the map. For example, when the map 800 is reduced, the mesh size may be aggregated into 500m square and the prediction result 806 for each mesh may be displayed. . In this case, the predicted value may be the final predicted value, for example, by averaging the predicted values of the 100 m square mesh included in the 500 m square mesh.

As for the matter selection button 804, a plurality of matter types may be selected. In that case, for example, the predicted value of the deterrent effect amount for each case type may be summed up for each mesh and represented by color shading. In that case, for example, when the mouse cursor is placed on the mesh, the predicted value for each case type may be displayed in a balloon. In addition to the amount of deterrence effect, the incident probability calculated in the derivation process may be visualized and confirmed on a separate screen.

According to the first embodiment, each police officer can confirm through the computer 130 the location where the police are highly effective in deterring the occurrence of incidents, and can determine the destination of the police based on that information. As a result, it is possible to prevent the occurrence of incidents by carrying out preventive measures by carrying out security guards in places where the deterrent effect is expected to be high based on the relationship between the past record of security guards and the occurrence of incidents.

A system for predicting the effect of deterring the occurrence of incidents according to the second embodiment will be described. The second embodiment is a method of considering the warning priority for determining the warning destination when the deterrent effect amount for each mesh in the first embodiment is calculated for each case type.

FIG. 9 is a case priority weight table in which a priority weight is set for each case type. The user sets the priority of each matter type as a weight. For example, in the example of FIG. 9, a high weight is set for "murder", which means that the priority of "murder" is higher than other case types.

A prediction server according to the second embodiment will be described with reference to FIG. The prediction server of the second embodiment is obtained by adding a case priority weight storage unit 1000 to the prediction server 120 of the first embodiment (see FIG. 2(b)).

A case priority weight table is stored in the case priority weight storage unit 1000, and the case priority weight table is called by the prediction processing unit 213 after the amount of deterrence effect for each case type is calculated.

A process of predicting the effect of deterring the occurrence of incidents by machine learning executed by the prediction server 120 will be described with reference to the flowchart of FIG. 11 for prediction. FIG. 11 is a flowchart for calculating the warning priority for each mesh by adding new processing to the first embodiment.

Only parts that are changed from the first embodiment will be described. First, the request receiving unit 215 acquires the case priority weights together with the input parameters from the user, and stores them in the case priority weight storage unit 1000 in the form of a case priority weight table (S1100). After the prediction processing unit 213 calculates the deterrence effect amount for each case type, the case priority weight table is obtained from the case priority weight storage unit 1000 .

Then, the deterrence effect amount for each case type is multiplied by the priority weight of the corresponding case type in the case priority table. The result of multiplying the deterrent effect amount and the priority weight is defined as the alert priority for each mesh (S1101).
Finally, the output unit 214 saves and outputs the warning priority (S1102).

Thus, according to the second embodiment, by multiplying the case priority weight, for example, when there are two meshes with the same deterrent effect amount, the one with the more important case type can be presented with priority. can be done. Also, for example, when the deterrence effect amount for multiple incident types is calculated for each mesh, if the result of multiplying each by the incident priority weight is summed, it can be regarded as the overall priority and alerted. You can decide where to go.

An incident prevention effect prediction system of the third embodiment will be described. The third embodiment expands the deterrent effect amount calculation according to the presence or absence of the implementation of the alarm in the first embodiment to the deterrent effect amount calculation according to the difference in the embodiment of the alarm. Here, the embodiment of the patrol means information defined based on the implementation means of the patrol, such as the number of police officers at the time of patrol and whether or not a vehicle is used.

FIG. 15 is a diagram showing an example of an embodiment pattern of police. Based on the information contained in the GPS database 103, combinations of the number of police officers and the presence or absence of vehicle use are defined as embodiment patterns, and code values are assigned. If other embodiment patterns can be defined from the GPS database 103, patterns other than the number of guards and vehicle use may be defined. It should be noted that the type of warning also includes no execution of warning, and the code value is set to 0 here.

FIG. 12 shows that the warning variable in the learning input table of the first embodiment is changed to the warning type variable. In the warning variable, the presence or absence of execution of the warning is represented by a binary value, but the warning type variable is represented by a code value defined in the embodiment pattern of the warning. For each mesh ID, date, and time zone in the past, the embodiment pattern of the police actually performed at that time and place is set as the value of the warning type variable. The embodiment pattern is included in the feature amount as a warning type variable, and the relationship with incident occurrence is learned.

Next, with reference to the prediction flowchart of FIG. 13, the calculation of the deterrent effect amount based on the difference in the types of alerts will be described. Here, FIG. 13 is a flowchart for predicting the deterrent effect amount for each alert type. Only parts that are changed from the first embodiment will be described.

The request receiving unit 215 acquires the alert resource based on the above embodiment pattern from the user (S1300). The police resource refers to the number of people, the number of vehicles, etc. that the user can assign to the police in the date, time period, and geographic area to be predicted. For example, if the embodiment pattern is defined as a combination of the number of people and the use or non-use of vehicles, the user inputs the number of people and the number of vehicles as a warning resource.

Subsequently, the data processing unit 211 determines an executable embodiment pattern from the alarm resource, creates an alarm type variable having each embodiment pattern as a value, and makes a prediction input table for executing a specific alarm type. is created (S1301).

As an example, consider a case where the police resources are two police officers and one vehicle. There are five viable alarm types as shown in FIG. As in the first embodiment, five prediction input tables are created in which only the values of the warning type variables are changed. The prediction processing unit 213 inputs the five prediction input tables into the prediction model, and calculates the incident probability. In other words, these are predictions of how likely an incident will occur when each type of security is implemented.

As for the deterrent effect amount, as in the first embodiment, the probability of occurrence of an incident occurs when the patrol type variable is "no patrol execution" and other specific values (for example, "two police officers and vehicles"). Calculate the difference between
As a result, the amount of deterrence effect when a specific police type (in the above example, "two policemen and vehicles") is implemented is calculated. This is calculated for each alarm type value (S1302).

In the third embodiment, if the amount of deterrence effect can be predicted for each type of patrol, for example, in the same mesh, if it can be determined from the prediction result that the deterrence effect will not change even if the number of police officers serving as patrol is increased, patrol resources can be saved. This allows for effective police enforcement.

Thus, in the above-described embodiment, by learning the relationship between past crime occurrences and the execution of police by the police, the effect of deterring the occurrence of incidents by the police in the place where the police will be executed from now on is predicted, and the effect of deterring the occurrence of incidents is high. Can provide location. That is, by predicting the effect of deterring the occurrence of incidents by the police for each location, it is possible to determine the effective destination of the guard based on the magnitude of the effect of deterring the occurrence of incidents.

100 data server 101 event database 102 geographic information database 103 GPS database 104 map database 110 learning server 120 prediction server 130 computer (user terminal)

Claims (10)

a data server having a case database that stores information on the date and time of occurrence of a case, location, and type of case, a geographic information database that stores geographic information, a GPS database that stores location information of police officers, and a map database that stores map information;
a learning server that learns information stored in the event database, the geographic information database, the GPS database, and the map database to build a predictive model for occurrence of the event;
a prediction server that predicts a deterrent effect amount of the occurrence of the proposal by referring to the prediction model;
a user terminal that visualizes and displays the deterrent effect amount predicted by the prediction server on a map by referring to input data input by a user;
The learning server is
By learning the geographic information and the location information of the police officers using the presence or absence of the occurrence of the incident as teacher data, learning the relationship between the past geographic information and police implementation status and the occurrence of the incident incident, constructing said predictive model of the occurrence of the proposal;
The prediction server is
Cases and implementations of vigilance at a future date and time using the prediction model constructed by the learning server and information on the date and time of occurrence of the proposal and the type of the proposal as the input data input from the user terminal. Predict the probability of occurrence of the incident in each case, and consider the difference in the probability of occurrence of the incident caused by the presence or absence of the police as the deterrent effect of the police officer's warning, and calculate the amount of deterrent effect of the occurrence of the incident seek,
The user terminal is
The map information acquired from the map database is displayed on the screen, and the deterrent effect amount of the occurrence of the plan output from the prediction server is visualized and displayed on the screen together with the map information. A system that predicts the effect of deterring the occurrence of incidents. The learning server is
Having a learning input table that holds mesh IDs of meshes, dates, time zones, geographic information variables, police variables, and incident variables,
The mesh is configured by meshing the prediction target geographic range,
The geographic information variable is information that associates the mesh ID with the date or the time zone,
The case variable is teacher data representing whether or not the case has occurred in the mesh ID, the date, and the time period. Aggregated from information on the date and time of occurrence and location,
The police variable is a variable that indicates whether or not the police was implemented before the occurrence of the proposal in the mesh ID, the date, and the time zone. 2. The system for predicting the effect of deterring the occurrence of incidents according to claim 1, wherein the system is aggregated from information. The prediction server is
having a prediction input table that holds the mesh ID, date, time zone, geographic information variables and alert variables of the mesh;
The mesh is configured by meshing the prediction target geographic range,
The geographic information variable is information that associates the mesh ID with the date or the time zone,
The police variable is a variable that indicates whether or not the police was implemented before the occurrence of the proposal in the mesh ID, the date, and the time zone. 2. The system for predicting the effect of deterring the occurrence of incidents according to claim 1, wherein the system is aggregated from information. The prediction server is
creating two types of the prediction input table, one for when the warning is executed and one for when the warning is not executed;
4. The system for predicting the effect of deterring the occurrence of incidents according to claim 3, wherein the amount of effect of deterring the occurrence of the plan is calculated by inputting the two types of input tables for prediction into the prediction model. The prediction server is
Obtaining the deterrent effect amount for each mesh,
The user terminal is
5. The incident deterrence effect prediction system according to claim 4, wherein the deterrence effect amount is displayed on the map for each mesh. The user terminal is
the map information obtained from the map database;
a date selection button for selecting a date for predicting the amount of deterrent effect of the occurrence of the incident;
a time zone selection button for selecting a time zone for predicting the amount of deterrent effect of the occurrence of the incident;
Display the prediction display button and ,
The prediction server is
In response to pressing of the prediction display button, the date selected by the date selection button and the time period selected by the time period selection button are sent to the prediction server;
The prediction server is
Obtaining the deterrent effect amount in the mesh using the prediction input table , the date selected by the date selection button, and the time period selected by the time period selection button,
The user terminal is
6. The incident deterrence effect prediction system according to claim 5, wherein the deterrence effect amount for each mesh is displayed on the map. The user terminal is
7. The incident deterrence effect prediction system according to claim 6, wherein the deterrence effect amount for each of the meshes is expressed and displayed on the map in shades of color. The prediction server is
further having a case priority weight table in which case priority weights for each case type are set;
Obtaining the result of multiplying the deterrence effect amount for each of the case types by the proposal priority weight of the corresponding proposal priority weight table as a warning priority,
The user terminal is
2. A system for predicting the effect of deterring the occurrence of incidents according to claim 1, wherein the warning priority is displayed. The prediction server is
Obtaining the deterrent effect amount for each type of guard when a specific type of guard is performed,
The user terminal is
2. The system of claim 1, wherein the amount of deterrence effect is displayed for each type of alarm. The prediction server is
10. The method according to claim 9, wherein, as the police type, a combination of the number of police officers and whether or not a police vehicle is used is defined as a police embodiment pattern based on the location information of the police officers contained in the GPS database. Incident occurrence deterrence effect prediction system.

Images