Hotspot Analysis

Tools: Display XY Data, Add Field/Calculate Field, Merge, Definition Query, Average Nearest Neighbor, Mean Center, Standard Distance, Directional Distribution, Spatial Join, Optimized Hotspot Analysis

Objective: the purpose of this project was to understand the spatio-temporal variability of crime occurrences in Austin, Texas by performing spatial analyses such as nearest neighbor analysis, mean center, standard distance, and directional distribution. I also use spatial joins to calculate crime density and conduct an optimized hot spot analysis to identify statistically significant crime clusters.  

For this project, I specifically conducted analysis on theft and aggravated assault  data.

Average Nearest Neighbor, Mean Center, Standard Distance, and Directional Distribution

Findings: 

• Average Nearest Neighbor: Measures how clustered or dispersed points are by comparing the observed mean distance between points to the expected mean distance.

  • The ANN for theft ranged from 0.34 to 0.38 and the ANN for aggravated assault ranged from 0.51 to 0.55, indicating clustering.

• Mean Center: The geographic center of a set of points.

  • Based on the mean centers, the most of these crime incidents occurred in downtown Austin - northwest, north, and east of the UT Austin campus. 

• Standard Distance: Measures the spread of points around the mean center, indicating how dispersed the incidents are.

  • The standard distances for theft and aggravated assault was around 0.093 and 0.098  decimal degrees respectively, or around a 5.5 mile spread.

• Directional Distribution: Shows the direction and spread of points, capturing the dominant orientation of the crime incidents.

• The directional distributions for both crimes tell us there is a north-south pattern in the same direction as major highways, such as I-35 and Mopac expressway. 

Crime Density

I created crime density maps by joining the Austin crime data to the census tracts shapefile. I then created a crime density field, calculated the crime density for each census tract (crime count / population), and symbolized the layer.