Data Acquisition:
In order to examine the spatial and temporal clustering of commercial robberies in Toronto, Ontario, data were acquired from the following sources:
Open Data, Public Safety Data Portal, Toronto Police Service
- Robberies, 2014 to 2018 (ROBBERY)
- Time of Occurrence
- Reported Day
- Reported Month
- Parsed into: COMMERCIAL, RESIDENTIAL, and OUTSIDE
Census 2016, Statistics Canada
- All roads (ALL_ROADS)
- Population aged 15 and above, by dissemination area (acquired via CHASS Data Centre) (POP15)
- City of Toronto outline (OUTLINE)
- City of Toronto dissemination area (DA)
- City of Toronto land use (LAND_USE)
- Major roads (ROADS)
- Light Gray Canvas
Data Cleansing:
1) Time of Occurrence (TIME): The original format (yyyy-mm-dd hh:mm:ss) was reformatted to hh:mm, using a 24-hour time convention. The attribute data for Robberies, 2014 to 2018, was converted to an excel file (Conversion Tools > Table to Excel), where appropriate edits were made in excel. The edited excel file was converted to a table (Conversion Tools > Excel to Table) and then joined with the Robberies, 2014 to 2018, attribute table (Joins and Relates > Join) using the FIDs. Lastly, this layer was exported (Data > Export Data) as an output feature class (.shp).
2) Robberies, 2014 to 2018 (ROBBERY): Data was parsed into commercial, residential, and outside Robberies according to their description in the attribute table. The following formulas were used parse out the 3 types (Select by Attributes): (1) “premisetype” = ‘Commercial’ OR “offence” = ‘Robbery – Business’ for commercial robbery; (2)”premisetype” = ‘Apartment’ OR “premisetype” = ‘House’ for residential robbery; and (3) “premisetype” = ‘Outside’ for outside robbery. For each selection, a layer was created (Selection > Create Layer From Selected Features) which was then exported (Data > Export Data) as an output feature class (.shp).
3) Population Aged 15 and Above (POP15): This data was provided in an excel file format which had to be matched to appropriate dissemination area (DA) to be properly appended. The DA attribute table was converted to an excel file (Conversion Tools > Table to Excel) in order to match the DAUIDs for the 2 files while also ensuring the population data had appropriate FIDs. The matched population data was converted to a table (Conversion Tools > Excel to Table) and joined with the DA attribute table using the FIDs (Joins and Relates > Join). Lastly, this layer was exported (Data > Export Data) as an output feature class (.shp).
Note: There were 14 DAs with no values reported for population. While efforts were made to interpolate the value by performing a polygon-to-polygon prediction (using Areal Interpolation in the Geostatistical Wizard) as a Poisson distribution (i.e., event counts), our efforts to build a valid model by adjusting the variography were unsuccessful due to time. More specifically, changing the model type from Spherical to K-Bessel proved to be quite time-consuming (i.e., waited over an hour with incredible computer lag which eventually shutdown the software). As a result, we examined the underlying land use in many of the DAs with missing values and found that large portions were classified as ‘Government and Institutional’. After looking up locations of Correctional Services Canada institutions within the City of Toronto and then matching it to the approximate location of the DAs with missing values, a value of ‘0’ were assigned to these DAs as a result.
4) City of Toronto Dissemination Area (DA): As this layer was initially acquired as a polygon feature class, it had to be converted to a point feature class in order to use in CrimeStat. Thus, points were created at centroids of each polygon (i.e., each DA) to convert the data into a point feature class (Data Management > Features > Features to Point (inside)).
5) X,Y Coordinates: Geographic coordinates in the form of x,y coordinates were added to COMMERCIAL, RESIDENTIAL, OUTSIDE, and DA_CENTROID shapefiles to be used in CrimeStat.