Module 1: Crime Analysis

In this week's module, we delved into the realm of spatial analysis tools, exploring their application in depicting crime hotspots within the Chicago and Washington DC regions. In the first portion of our lab, we undertook spatial joins between distinct feature classes, effectively generating choropleth maps that elucidated areas with elevated burglary rates.  For Washington DC, we conducted spatial joins between two feature classes to create choropleth maps showcasing where there were high burglary rates. The second portion of our lab involved employing three different techniques to identify and visualize hotspots. The second portion of the lab had us use three different techniques to identify hotspots. By successfully using these three techniques in conjunction with appropriate symbology, we completed the goal of detecting and visually representing crime hotspots. 

The first technique used was Grid-Based Thematic Mapping. A spatial join was done between the grid cells of Chicago and the 2017 homicides resulting in a new feature class that had a new column depicting the count of homicides per grid cell. This provided the needed attribute field to find the grid cells with the highest number of homicides. After exporting the first 62 rows containing the highest number of homicides per grid cell, I used the dissolve tool to create a multipart feature with a smooth service. The results are seen below. 


The second technique was Kernel Density. Similar to the previous technique, a spatial join was done between two feature classes, creating the needed feature class of total homicides in each grid cell. The Kernel Density tool was used to turn the feature class into a raster. I edited the symbology to show only two break values of the mean and the mean times 3. Once this was completed, the Raster to Polygon tool was used to create a new feature class of the raster and enabled the selection of values equal to 2 which signified the hotspots of homicides. The results are below. 


The last technique used was Local Moran's I. Like before, a spatial join was done between the census tracts and 2017 homicides. With this new feature class, I used the Join Count column and divided it by the total households column multiplied by 1000 giving me the crime rate. Then I searched for the Cluster and Outlier Analysis tool inputting the feature class from the spatial join and using the crime rate to find the different levels of crime from low to high-high. To spotlight the highest level of clusters, I queried the high-high clusters and exported them into their own shapefile. To finish, I used the dissolve tool with the COType IDW field. The results are below. 

 
In summary, this laboratory exercise provided me with insights into employing a range of techniques for hotspot identification and determining the most suitable approach for crime analysis. While encountering some challenges with the Field Calculator for Local Moran's I, the issues stemmed from utilizing an incorrect column with a different data type compared to the other. By leveraging these techniques, it becomes possible to conduct crime analysis based on historical data, facilitating the prediction of future areas warranting heightened policing efforts. Ultimately, I discovered that Kernel Density emerged as the most effective method for pinpointing crime hotspots and anticipating future regions requiring greater policing attention. 


Comments

Popular Posts