Crime data can have spatial dimension embedded in it, hence spatial analysis and crime mapping can be very effective in understanding criminal phenomena. Crime data are considered as spatial crime data when the unit of analysis are geographically referenced (Bruce & Smith, 2011). Since crime events have inherent geographical properties in it, it is assumed that these events are not randomly distributed across space, rather these events are these events are clustered across space (Yavuz & Tecim, 2013). In this research, the focus has been on the assault crime. The World Intellectual Property Organization (WIPO) defines assault as the crime, where the offender willfully and unlawfully attempts to force or be violent to another person (WIPO, 1960). As analytical method geographically weighted regression (GRW) method has been used. The core concept behind the GRW is that the relationship between the dependent and independent variables vary across the space, instead of being same across the geographic space and it takes spatial non-stationarity into account by computing many local regression in the sample (Brunsdon, Fotheringham, & Charlton, 1996). GWR is calculated as
Where the parameter βoi(u) describes a relationship around location u and is specific to that location. Prediction for the dependent variable yi(u) can be made if measurements for the independent variables are also available at the location u (Charlton, Fotheringham, & Brunsdon, 2009). The objective is to find out the set of explanatory variables that can explain the assault crime in the city of Calgary for the year of 2006 using GWR method.
In this assignment, GWR method was applied to find out the set of explanatory variables of the assault crime. The results reveal that GENDER. AGE30.39, MEDIAN.FINC, Population variables can predict the assault crime with a R2 value of .61. Using R2, AIC and CV it was found out that GWR is a better model than Global regression model. Population variable was highly correlated with dependent assault crime variable. However, the results of this model can be biased as the residuals were not entirely normally distributed.