Crime Assortativity Analysis of Coterminous Wards

We present three-levels of assortativity analysis to assess the crime similarity of coterminous wards (that share a common border). We use the crime data for the City of Chicago to showcase our approach. We first conduct principal component analysis on a ward-crime type dataset featuring the number of occurrences of a crime type in a ward. We compute a weighted average PC_crime_score for each ward based on the entries for the ward in the high-variance principal components and their variances as weights. We categorize a ward as crime hotspot if its PC_crime_score is positive. The high-level of assortativity analysis involves computing the Kendall’s concordance correlation coefficient for coterminous wards based on their hotspot classification (a binary metric). The mid and low-level assorativity analysis respectively involve computing the Pearson’s correlation coefficient for the coterminous wards based on their PC_crime_scores and number of occurrences of the individual crime types in the wards.