Abstract: The vast amounts of data that are generated and collected in today’s world bear ...
Expand
Abstract: The vast amounts of data that are generated and collected in today’s world bear immense potential for businesses and authorities. Innovative companies already adopt novel analytics methods driven by competition and the urge of constantly gaining new insights into business operations, customer preferences, and strategic decision making. Nonetheless, local authorities have been slow to embrace the opportunities enabled by data analytics. In this paper, we demonstrate and discuss how latent structures unveil valuable information on an aspect of public life and communities we all face: criminal activity. On city-scale, we analyze the spatial correspondence of recorded crime to its physical environment, the public presence, and the demographical structure in its vicinity. Our results show that Big Data in fact is able to identify and quantify the main spatial drivers of criminal activity. At the same time, we are able to maintain interpretability by design, which ultimately allows deep informational insights.
Collapse
Semantic filters:
projection to latent structure
Topics:
criminality analytical information system social media Twitter sensemaking
Methods:
partial least squares regression cluster analysis partial least squares path modeling archival research Breusch-Pagan test
