Julia Chen
Understanding and predicting the time, location, and type of crime in our cities has become more possible as large amounts of data are available, and the addition of artificial intelligence (AI) in recent times. This study examines crime incident data collected from the City of Chicago from 2001 to September 2023 in an effort to understand and predict crime patterns. This study concentrates on patterns in location, time, and type of crime that may help police target their resources where they are needed most and lead to a better understanding of crime patterns in their city. Features in the dataset include the date of incident, primary type of crime, description, police district, and location where the incident occurred.
This study focuses on the following areas: generating strong association rules for crime incidents, using models to predict if an arrest will be made given other features in the dataset, and investigate the most common characteristics of crime incidents.
In the City of Chicago, crime incidents are most likely to occur in July and August, on the 1st of the month, and noon through night hours. Crime incidents are least likely to occur in February and December, the 31st day of the month, and early morning hours. The most common crime types are theft and battery. Association rules show narcotics is commonly occurring with arrest, which is further supported by decision tree model results when predicting arrest.
The results of this study will inform police districts in the City of Chicago when and where to deploy resources for crime prevention and response. The classification results also allow police districts to see which types of incidents have successfully led to an arrest and where they can improve. The data mining techniques described in this study may be applied to a visualization system where police and city administrators would have the ability to explore the data in more detail and automatically import the most recent data available.