LA Public vs. Private Crime

Members: Elias Demko, Catalina Flores, Joshua Harvey, Leo Kayatsky

          This project focused on the Los Angeles Crime Dataset. This dataset is a record that organized detailed descriptions of reported crimes over the last five years, including areas, dates, demographics, and specific offenses. Our goal was to explore how the setting of a crime, especially the distinction between public and private property, influences the types of offenses committed, with an interest in the role of perceived privacy and surveillance. The main questions for our research were: How does the setting impact or influence the crime that has taken place? And, what is the difference between crimes that occur on public property versus on private property?
          A large step in the project was processing the raw LA Crime Dataset, which contained 1,005,092 rows. Each row, representing a crime incident, included a Premise Description field. To separate public and private settings, we manually reviewed and categorized the premise descriptions based if a location was primarily a place where somebody could live or not.
          The project involved several stages of work. After categorizing premise descriptions into public and private we made several charts to break down the places with the most reported crime and then what types of crimes were most common in that location. This included analyzing crimes in residential settings, with a detailed breakdown of offenses in "Single Family Dwelling". We did the same process for outdoor/public crimes, focusing on "Street" locations.
          Our analysis of the dataset showed crime profiles for public versus private settings, suggesting that location and perceived surveillance do influence criminal activity. In private settings, such as single-family dwellings, there was a higher incidence of crimes that appear to utilize privacy and reduced oversight. "Theft of Identity" was the most common (26% of crimes in single-family dwellings), followed by "Burglary" (15%), and "Intimate Partner - Simple Assault" (13%). This pattern suggests that the secrecy created by private spaces may be promoting these offenses. In public settings, most commonly, crimes happened on the street. These were opportunistic crimes, especially those targeting vehicles, with "Vehicle - Stolen" being the most frequent at 40%. Poirier's deconstructive reading helped us question the public/private labels of crime scenes by exploring what "absent meanings", such as specific surveillance conditions like lighting or CCTV, were not shown within broad data categories. Koopman’s format anatomies helped us conclude that the dataset's premise descriptions directly limited our ability to analyze the varying levels of surveillance associated with each crime.
          Perceived surveillance, while not outright measured in the dataset, appears to be a significant factor. The types of crimes in private settings suggest a the idea of reduced risk due to a lower sense of being observed. In contrast, the nature of surveillance in public spaces creates different kinds of opportunities. Location is an active variable in shaping crime patterns, and the very structure of the data, is important for our understanding. These findings have implications for urban planning, crime prevention strategies, and the ongoing debate on privacy and security in society.