Skip to content

Latest commit

 

History

History
24 lines (18 loc) · 4.97 KB

File metadata and controls

24 lines (18 loc) · 4.97 KB

Analysis of Crime Rates in Italy from 2006 to 2021

Introduction

Crime is an ever-present topic: every day, in all kinds of media, we are constantly bombarded with news and reports related to all kinds of crimes and criminals. The issue of crime has been a longstanding concern in Italy, impacting various aspects of society and posing significant challenges to law enforcement agencies and the overall well-being of its citizens. According to the report regarding the perception of safety published by ISTAT in 2018: "In 2015-2016, it is estimated that 27.6% of citizens consider themselves little or not at all safe going out alone at night, for 38.2% the fear of crime influences their habits a lot or enough. [...] Women's sense of insecurity is significantly higher than men's: 36.6% do not go out at night out of fear (compared to 8.5% of men), 35.3% when going out alone at night do not feel safe (19.3% of men)." Continuing to study the report, a general improvement in feelings of worry over the years can be noticed. Nevertheless: "33.9% of citizens believe they live in an area at risk of crime (very or fairly)". Also: "There is a widespread opinion that the police should be on the streets more often (55.5%), be more numerous (44.2%) or more present in the area (26.6%), and particularly in areas at risk (20.5%) and at night (20.3%)".

The aim of this project is to thoroughly investigate the trend of crime indices in Italy from 2006 to 2021, and try to answer the following social questions:

  • Is this general feeling of concern and danger, constantly fed by politicians and the media, well-founded?;
  • What is the general crime trend observed over the years on our territory?.

This work is proposed as a quantitative research based on secondary data. Multiple datasets were combined, all downloaded (and downloadable for everyone) through the ISTAT web application at the following links:

Methodology

Once the data was obtained, the cleaning and analysis work was done entirely using Python libraries. In particular: Pandas and NumPy were used for data manipulation and analysis while Streamlit and Plotly were used to create interactive dashboards for data visualisation (including one obtained with the help of shapefiles of Italian provinces). All data manipulation and analysis work can be observed in detail, with explanations, in the Jupyter Notebook file called “data_manipulation.ipynb” located in the repo.

The variables correspond to 54 diverse types of crimes (other than these, there are two more variables: one is listed as “others” and contains multiple crimes not previously considered within it; the other represents the “total” amount) committed from 2006 to 2021 (a time span of 16 years). Furthermore, to try to limit what I would call the population bias (larger units of analysis tend to have more crimes than smaller units of analysis) a metric was manually introduced to measure the incidence of the values per 100'000 inhabitants. This introduction required the addition of data concerning the resident population, variable that was absent in the initial dataset.

The web applications can be seen at the following links:

As can be seen by observing the interactive dashboards built with Streamlit, the possible conclusions that can be drawn from this project are almost unlimited. Nonetheless, some conclusions have been drawn as a demonstration in the "results.ipynb" Jupyter Notebook. The work was performed on 4 different units of analysis: national level, macro-area level (with a focus on North-East), regional level (with a focus on Emilia Romagna) and, lastly, provincial level (with a focus on Ferrara). The comparison among different levels of units of analysis leads to a better understanding of the problem, without the need for generalization. It is important to remember that relying solely on the data made available in the tables above is not the best approach to gaining a deeper understanding of the problem at hand. In order to get a more complete overall idea of the problem, the data in the tables should be supplemented with the visualisation tools made available. For example, some large reductions or huge increases (-100% or +100%, for instance) could most likely concern crimes with few reports and which, therefore, do not pose a particular danger to the community. Other cases, on the other hand, might concern crimes that, yes, show a decrease compared to previous data, but which corresponds to an up-and-down trend over time rather than a real downward trend.