Methodology

There are several aspects to my methodology. These include data management, creating maps detailing crimes in neighbourhoods and community housing locations and police stations in neighbourhoods, Generalized Linear Regressions, Spatial Autocorrelation analysis and Hotspot Analysis.

Data Management:

  • Ensure that all the data is standardized. This is done by ensuring several things.
    • The correct time frame is chosen.
      • As the data from the “Major Crime Indicators” was only available from 2014 to 2019, I had to ensure that the data from “Homicides” was also from 2014-2019.
      • To do this, I selected by attributes and selected all the data in the “Homicides” dataset from 2014 to 2019 and created a new layer.
    • The data is all in the same projection.
      • As all the data came from the same source, the city of Toronto, the data is all in the same projection. This is the case as the city offers several projections for downloading the data, and I made sure that I chose the same one every time.
      • The one exception was the “Neighbourhood Improvement Areas” dataset, which was offered only in the WGS84 projection.
        • I used the “Project” tool to ensure that this dataset would be in the same projection as all the other data.
    • The correct data is chosen.
      • From the data available in the “Major Crime Indicators” dataset, I wanted to investigate break and enters and assaults.
      • I selected both break and enters and assaults by selecting by attributes and created a new layer for each crime indicator.

Creation of Maps detailing Crimes in Neighbourhoods

  • I took a layer with the points of crimes created in the last step and used the “Summarize Within” tool to input that data into each Neighbourhood.
  • I included the “Neighbourhood Improvement Areas” dataset.
  • I repeated this for the other two types of crimes.
  • I created 3 maps, one of each three types of crimes.

Creation of Maps detailing Community Housing and Police Stations in Neighbourhoods

  • I took the point data for both the Community Housing dataset and the Police Facility Location dataset and used the “Summarize Within” tool to incorporate it into the Neighbourhood dataset.
  • I created 2 maps, one for each type of variable.

Generalized Linear Regression (GLR)

  • I took all three maps detailing crimes in Neighbourhoods and used the “Summarize Within” tool to input the Community Housing, Transitional Housing and Police Facility Location datasets.
  • I ran the GLR using all three variables with the count of Points for the crimes as my dependent variable.
  • Due to a lack of data in the Transitional Housing dataset, the GLR did not function. As a result, I ran the GLR using the Community Housing and the Police Facility Locations as my exploratory variables, with the three different crimes being the dependent variable.

Spatial Autocorrelation

  • To run the Spatial Autocorrelation, I took the layers created in step one of the GLR section and used the spatial autocorrelation tool to determine how spatially correlated the three different types of crimes are.

Hot Spot Analysis (Getis-Ord Gi*)

  • To run the Hotspot Analysis, I took the layers created in step one of the GLR section and used the Hot Spot Analysis (Getis-Ord Gi*) tool to determine statistically significant hotspots and coldspots of the three different types of crimes when compared to the number of police stations and the number of community housing units.