As a Bronx resident I often wondered, how is it that a place like the Bronx, so close to the city, can stay as underpopulated as it has for as long as it has? This led me in a search of statistics about the only borough attached to the mainland. Originally when I decided to use the data from Lehman I did not know what to expect. In a way this was good because I had no preconceptions about what I would find. Meaning that I was looking at the data without bias; which would allow me to notice any patterns or trends that would emerge from the data (Few, 2009).
Graphing the data took less time than expected with the inclusion of the time it took to learn the basics of using CartoDB. Because the dataset included the zip codes of each section in the Bronx I was able to create the visualization without the use of shapefiles. The more challenging aspect was the actual analyzation of the data. At the start I looked through the data itself to see what information I wanted to show on the map. Finding numbers for both families and households made me go on a search to understand the difference between a family and a household. The difference was important because I intended to determine the population density by dividing either the households or families by the number of housing units. However, without knowing the difference between the two I could calculate the density incorrectly and thus give an inaccurate representation of the population density. I ended up using the number of households to calculate the ration of individuals living in each housing unit. The ratio of household to the housing units was encoded into the color of the Choropleth Bronx map. Darker colored areas in the map represented more densely populated areas while the lighter colored areas represented less density. Because the map was about population density I decided to include relevant information such as the number of families, the number of households, and the number of housing units. Such data would be shown when the user hovered over one of the zip code areas on the map.
I tried a multitude of color schemes before settling on the monochrome light yellow to dark blue scheme. The thought process being that the color blue was a more neutral color for the map. Using green would have been the next best choice but I thought viewers would associate the visualization with either the green areas or parks in the Bronx. Red was too bright and seemed to imply that there was something dangerous or imminent about the data, it would also pull the user’s attention immediately to the more dense areas; where I would want the user to take in the whole map instead (MacDonald, 1999). After creating the visualization I took some time to look at the story being told by the visualization. Four areas were less populated than all the others: Riverdale, Woodlawn, City Island and Hunts Point. Out of these four Hunts Point stood out as it was the only one in the southern part of the Bronx. This once again made me look through the data and I found that Hunts Point was also the only area of the four that had a higher than 50% rental property ratio. Meaning that out of the four areas only Hunts Point had more than half of its properties as rental properties. Another piece of information that I considered important but was missing from the data was that Hunts Point is an industrial area, home to many markets, a water treatment plant, auto shops, junk yards and warehouses but few housing units. Having found this information it made me realize that the number of rentals was important information for the users as it affected the population density significantly in the other three areas.
I continued to look for other patterns in the data specially among the areas that were less populated. The data revealed that at the same locations, excluding Hunts Point, the percentage of whites living in the area was higher than in any other area. In the areas where the percentage of white residents was highest, white residents outnumbered any other race in those areas. This led me to believe that race was somehow connected with the population density or the lack thereof. For that reason I decided to allow users to click on the areas in order to reveal a racial breakdown of the population. Having information delivered in two stages, once when the user hovers over an area and second when the user clicks for more details, makes sure that the user is not overwhelmed or being provided with too much information at first glance (Few, 2009). Allowing for the user to look at the map and develop their own understanding out of what the visualization is presenting.
The cultural human imprint could not be entirely removed from the map, because the map itself was about the Bronx (Sula, 2015). I decided to keep the borough’s boundary outline as well as the Bronx zip code boundary lines, since they were the basis for the visualization and how the data was organized. However the other boroughs around the Bronx were deleted and so were any other georeference items about the Bronx, including street names and their representations. I admit to wanting to find deeper meaning in the data, but ultimately the visualization ended up being very straight forward, a look at the population density of the Bronx.
References
Few, S. (2009). Now you see it: Simple visualization techniques for quantitative analysis (1st ed.). New York: Analytics Press.
Macdonald, L. (1999). Using color effectively in computer graphics. IEEE Computer Graphics and Applications, 20-35.
Sula, C. (2015, April 2). Mapping and Countermapping. Lecture conducted from Pratt Institute, New York.