Exploring Crime Patterns in Georgetown, Guyana: Leveraging Spatial Analysis and Machine Learning on Mass Media Data for Comprehensive Insights.

Abstract

The deluge of big data has opened up new research frontiers for the incorporation of both spatial and temporal dynamics of crime. However, regardless of the geographical attribute and detailed description of the phenomenon they provide, these data types are often unstructured, and therefore data mining requires a special approach. This ongoing research draws on newspaper data spanning over five years to explore the crime patterns in Guyana, a country that has experienced an increasing crime rate over the years. With the aim of understanding the crime landscape of this country and making a prediction for law enforcement, this research focuses on violent crimes and employs an innovative approach that integrates both spatial analysis and machine learning techniques from the data extraction phase to the gaining of comprehensive insight into the dynamics of crimes and prediction. The preliminary findings revealed that crime incidents are not random and there is indeed a distinct pattern where crimes are clustered around particular street segments. The study provides valuable insight to law enforcement agencies and policymakers on new ways to practice predictive policing and at the same time, strengthens the role of mass media in spatial analysis of crime.