Space probability assessment and analysis tools-application and Public Security Industry

Spatial probability assessment and analysis tools

1. Spatial probability assessment analysis description
Here, the space probability refers to a probability of a thing happening somewhere. If there is such a proposition:
A total of May cases occurred in the City in February 200, of which 16 occurred in Residential Area A. Are there any exceptions?
The exception described here is the case that occurs throughout the city. If there are no special conditions, the case should be evenly distributed across all corners of the city (of course not possible, this is related to many situations such as public security, floating population, educated class, and economic conditions. Here we will discuss the pure natural space measurement relationship, and we will not consider all kinds of human and social factors for the time being.) How many cases will happen in community a theoretically? What is the percentage of cases in the city? Compare the actual case data to determine whether an exception occurred.
As shown in:

 
The large yellow area indicates the whole city, the blue No. 1 and No. 2 indicate the two districts, and the black circle indicates the case.
If we do not perform spatial probability assessment and analysis, how many cases have occurred in the No. 1 region? Is it normal? If not, what is beyond the normal value?
The following uses our spatial probability analysis tool to analyze the situation:
2. Tool description
The spatial probability assessment and analysis tool is a function written using Python scripts. It is encapsulated into tools through ArcGIS. the main interface is as follows:

 

There are a total of seven parameters, two of which are used to store temporary data settings, at least two layers need to be entered (where, the Space Constraint range layer and the analysis area layer may be the same layer .)
The seven parameters are described as follows:
Space Constraint range layer: the largest analysis area used to calculate the total magnitude of the area. It must contain the complete set of all the analysis areas.
Analysis area layer: the area where the probability of a space is to be analyzed. It must be a subset of the Spatial Constraints.
Event data point layer: the layer of the event point actually used for analysis.
Area analysis result output layer: the output layer is used to output the final analysis result. The structure is consistent with the analysis area layer.
Analysis Field: an immutable unique identifier in the analysis area layer (note that it cannot be a variable identifier such as objectid ).
Temporary output result storage space: the storage space of the data temporarily generated during the analysis process.
Temporary statistical result table: a temporary result table generated during analysis. Note that the result table cannot be placed in the element dataset. Therefore, you must set a parameter independently.
Total Space random rate: Total number of samples used for Reference Probability. The default value is 10000. According to the statistical principle, the more samples, the closer they are to the theoretical probability. However, the larger the total number of samples, the larger the calculation amount and the longer the consumption time.

 

3. tool usage
 
Running content:
 
Two sets of results are calculated during the operation. One is based on the total number of samples, and the other is based on the actual sample data to obtain the actual probability.
The above data will also be written into the output result.

From the results, we can see that in Region 1, when the total number of samples is 10000, the number of theoretically occurring in this region is only 149, and the spatial probability is only 1.49%, in fact, 12 of the 102 cases occurred in this region, and the spatial probability is as high as 11.76%, exceeding the theoretical probability by nearly 10 times. In this way, we can think that a serious incident occurred in this region.
In theory, the Area No. 2 is 13%, which is actually 17%, slightly higher than the theoretical data. We think this is a normal situation.

4. instance demonstration
 
For example:
Jurisdiction is the area chart of a certain area of a city.
Crimepoint_all is the spatial distribution of all cases in this region.
Now, we need to analyze that in these neighborhoods, the cases in those neighborhoods are higher than the theoretical ones.
The analysis process is as follows:
 
The running process is as follows:
 
The running result is as follows:
 
Through graphic rendering, the area lower than the theoretical value is set to blue, the area higher than the theoretical value is set to red, and the following spatial distribution is obtained:
 
We can see that the difference between our theoretical probability and actual probability is from-0.078 to 0.071. The distribution is as follows:
 

The two most challenging areas are 92 and 39 ., As follows:
 

Based on the calculated results, we can continue to take the corresponding analysis. Why is the probability of this region so high? Is there any special situation? If we use the time-based analysis throughout the year, it will produce more information.
5. Subsequent instructions
This tool uses random spatial distribution to calculate the theoretical probability without considering human factors. Therefore, the spatial model is very simple. If you can add various related factors to the tool in the future, the analysis results are more accurate.
Here is just a thought of spatial probability assessment, which is still very imperfect. We hope that interested colleagues can come up with valuable ideas to complete this model.
By: ESRI _ prawns Lu

 

PS: if you are interested in the above tools, you can follow Sina Weibo and @ ESRI _ prawns Lu. You can leave a private email to ask for it. (Appendix: The script used by this tool isArcGIS 1, 10.1If you need to run the program, install ArcGIS 10.1)