In 2012, the United Nations Office on Drugs and Crime (UNODC) released intentional homicide statistics. UNODC Homicide Statistics 2012 has over 1,000 records over a seventeen year span, from 1995 to 2011. The data was collected from over twenty sources, which are categorized by being either from criminal justice or public health systems. Data was collected from over 200 countries or areas.
The main question examined throughout the visualization of the data was how the number of intentional homicides fluctuated throughout the years, while looking at the data as a whole, by country or area, and by year. It was also important that the rate of homicides be examined against the number of homicides.
The following visualization examples influenced many of the design decisions taken to throughout. These examples are only a reflection of the initial thought process, which evolved overtime and is reflected by the final visualization.
Choropleth Map: With the inclusion of geographical location in the data, it was important that a visualization of each country or area was provided, but a simple map would not be enough. A choropleth map would not only provide a visualization for the location of each country or area of the recorded homicides, but the shading would allow the user to correlate the number of homicides for each of the represented country or area, darkest being the highest. The final choropleth map can be filtered by country or area, number of homicides, source type, and year.
Bar Chart: Initially a bar graph was considered to visualize the countries or areas and the overall number of homicides. Similar to this example, each geographical location, on the horizontal axis, would be assigned a color. With over 200 locations in the data, the bar graph was not the best option, even after placing the locations on the vertical axis instead. Including filters provided a temporary solution, in terms looking for numbers from countries or areas. However, it was too overwhelming to provide an overview, visually.
Treemap: Although not shown in the final visualization, a treemap was initially considered to provide users with a visual of the overall number of homicides per year. The larger and darker the rectangle, the higher the number of homicides. Filters were applied that would provide users to see the treemap by country or area, and source type.
The dataset was made available by UNdata, a data access system to United Nations databases. There were several download format options, but it was downloaded to Microsoft Excel as a commas separated values (csv) file. The dataset was opened in Tableau Desktop, and the completed dashboard was uploaded to Tableau Public.
A dashboard was created with an interactive filled choropleth map of each country or area, and two line graphs that show, per year, the number of homicides and the rate of homicides per 100,000 population, respectively. With the nature of the data, various shades of red was used throughout the visualization, with the exception of the rate line graph, which is orange because it would decrease the amount of stress for the user to distinguish between the two line graphs. The entire dashboard can be filtered by the same filter controls, which will provide the user with a complete story throughout their navigation of the dashboard.
After speaking with users, it was decided to focus the visualization on the information it is currently presenting (count, country or area, rate, and year of intentional homicides with filtering made available by the source type). This allowed the questions addressed throughout the process to be fully realized. This also resulted in the exclusion of the sources data within the dataset. The sources stated where each of the records came from, which was initially visualized through a descending bar chart. Each source was placed on the vertical axis, while the years were placed on the horizontal axis.
It would also be interesting to animate the map to show the fluctuation of the number of homicides throughout the sixteen years, worldwide and by country or area.
In the future, a story may be designed, in Tableau, that will provide an additional visualization of the various sources that each record was taken from. The previously mentioned bar graph can be used, which showed the sources of the homicide records in relation to the year. The bar chart could also be filtered by country or area, source type, and year.
Two vertically split choropleth maps can also be included. These maps would be similar to the current map, in that they will show the countries or areas shaded based on the number of homicides, but each map would be visualizing one of the two source types: criminal justice or public health.
Adding a dashboard, to the overall story, that focuses on the sources and source types from the data may inspire questions about the relationships between the type of sources providing the records and the countries or areas from which the records originate.