Introduction

New York City Housing Authority (NYCHA) provides housing for low-income residents of New York City and handles Section 8 housing in the five boroughs. Based on a many articles about the conditions that NYCHA residents live in, it is clear that the city does not provide a high quality of life for those who are tenants in NYCHA housing. With this in mind, I decided to use this visualization lab to explore in more concrete ways the neighborhood resources available to NYCHA tenants.

Related Visualizations 

While I did not go into the mapping exercise with related visualizations in my mind, when I ultimately decided to focus on the relationship between NYCHA housing and grocery stores, I looked to a previous Pratt SI student Information Visualization, J.E. Molly Seegers’, work (see below). Seegers’ visualizations did not attempt to find a relationship between housing developments and retail grocery stores, however, and seemed to only be exploring the retail grocery stores dataset.

Materials 

For this visualization, I used CARTO, an open source software that provides browser-based GIS and web mapping tools. I uploaded my previously downloaded datasets (CSV and GeoJSON files) to the CARTO site and styled the map directly on the web-based software. CARTO works very intuitively with any datasets that include latitude and longitude (or any other geographic data), so this was a relatively seamless process (challenges discussed in a later section). 

Methods/Process 

I downloaded all of my datasets from government sources, specifically New York City Open Data and the New York State Open Data. As mentioned, my primary dataset, the one that my map was to be based around, was that of all NYCHA housing. By partnering this dataset with one or two other datasets, I was hoping to reveal patterns surrounding quality of life or inequality in the areas with the greatest concentration of NYCHA housing. In addition to the NYCHA dataset, I uploaded datasets related to voting poll locations, open spaces/parks, grocery stores, and community districts. I experimented with several different analyses, including creating a choropleth map that indicated the number of NYCHA developments in a community district (see below) or the number of retail grocery stores in a community district as well as creating a buffer around retail grocery stores or NYCHA developments (in an attempt to discern distances between stores and housing). None of these analyses revealed any notable relationships. 

I encountered significant challenges with the dataset on existing retail grocery stores. Firstly, the dataset did not only include information for New York City, so I had to filter the dataset by the five counties (Kings, Richmond, Queens, New York, Bronx). I then had to geocode the dataset. The dataset did include latitude and longitude for many of the locations, but the points were together in one cell with the grocery stores’ address information. Therefore, with the help of Dr. Sula, I had to painstakingly separate the latitude and longitude into separate columns. I was then forced to delete a not insignificant chuck of the data that was unusable as it did not include geographic data. Despite the filters, geocoding, and deleting, I still ended up with a few data points in Portugal, which I had to manually remove from CARTO after uploading my geocoded dataset. 

Results 

I don’t find the result to be particularly visually appealing or easy to read, but if users spent time with the data points, clicking through to find out what food establishments exist around the NYCHA developments in their area, they may be able to learn something about the availability of healthy food to poor and working class people.

Reflections 

I found this visualization to be the most frustrating one that we have created so far. I ran into many challenges while trying to identify patterns or relationships in the datasets. Perhaps more time would allow me to better understand what these relationships might be. With my current map the sheer number of points on the map indicating retail grocery stores is so many that it is difficult to pull out any definite relationships. 

What I did notice as I was clicking through the retail grocery store data points was that many of the places listed as a “store” were in fact gas stations or bodegas rather than what one would typically consider a full-service grocery store—one with fresh, unprocessed foods available for purchase. Perhaps this is an issue with the dataset itself. Perhaps the dataset should indicate more specifically what kind of establishment each of the “stores” is, as they are not all the same. If one were to look at these data points without that knowledge, it looks as if there are no food deserts in New York City, which is certainly untrue. I myself recently lived on a block in Brooklyn with no full-service grocery stores within a mile or more (and four small NYCHA developments on the block!). Moving forward, I could either create such a dataset myself or focus in on a very small area of one borough to better analyze what kinds of resources (both food-related and otherwise) are available to NYCHA residents within a limited distance of their home.