The Crime Map of Israel, 2021
How to take open crime data and analyze it on a map?
The Israeli Police first opened its crime data to the public in Israel in 2011-to 2016 years.
Therefore, I had the urge to drill it down, interrogate the data, analyze it and visualize it on a map to tell a story.
After All, This Story and this important phenomenon impact our lives and the surrounding geography.
The Raw information is a CSV file that includes crimes per type police station related and the area where the offenses were committed, for 2011-2016.
The most detailed geography connection is the statistical area number.
I had decided to create a story and visual insights based on maps, not tables.
It’s Much more interesting, clear, and accessible.
So, the first step was to learn the data structure, values, and statistics. This was done easily after loading it to the PostgreSQL database.
Afterward, I started cleaning it from duplicate and null values, especially on the statistical area-id column. (it can easily be done in other methods, for instance: Pandas python library with Jupyter notebook).
Now we can try and join it to the geography.
I filtered the data for the year 2021, downloaded the statistical areas shapefile layer which is also publically available, and joined it with the statistic number.
Done?
Not quite. checking the join numbers turns out a lot of data was out (using inner/outer joins) After some assumption and reading, it turns out that the statistics numbers in the police data are based on older year (Israel CBS are updating it each few years).
I had to search for all the layers in the CBS site to find which is the one, use it or use the conversion tables so the crime data will reflect the true geography on the map.
The maps were created after cleaning and processing the data presenting the crimes by locality, and statistical area (Geographic divisions within the locality into smaller areas).
As well as joining the data to neighborhoods in Tel Aviv, in visually appealing methods and tools such as deckgl.js for smooth and 3d effects on a browser.
The visual methods used were: thematic maps, coloring each area with color codes, and dark colors for a high number of crimes. while this is a traditional method, I wanted to compensate for the lack of more detailed data than the statistical area, so I used the ST_GeneratePoints in PostGIS to emulate points spread over the area by the number of crimes. This will make a visual effect, where points are very densified over small areas. This function can be done also in Qgis and ArcGIS (create random points).
Normalizing the crime numbers per 1,000 persons (based on CBS data) was also done to make a more reliable picture of the crime ratio with the population in it.
Those types of open data and processing greatly help to analyze phenomena such as crime. Learning where it had decreased or risen, which neighborhood was the highest number of crimes of a particular type, and much more.
It helps in a wide range of industries such as insurance, Real estate, construction, academic research, etc. It helps to analyze risks, in planning, utilizing AI, and more methods.
