Fairfax County Public Schools Food Distribution Analysis

In July 2020, in the summer between 9th and 10th grade, I saw a tweet that piqued my interest. The tweet was a celebration that the county I lived in had distributed over a million meals through their new Grab & Go Food Distribution Program. This program had been enacted because schools had shut down due to COVID-19, and many students could not receive breakfast and lunch. As a solution, my county, Fairfax County in Virginia, allowed students to pick up packaged meals from various distribution locations. Parents or students could drive to distribution locations, mainly bus stops and libraries, to pick up meals.

The tweet was interesting to me for two main reasons:

The Food Distribution Program started towards the end of March 2020. In only four months, it had distributed over a million meals. Even for a county as large as mine, with over 1.1 million residents, this number was impressive.
I wondered how the program’s organizers knew how many meals were needed for each distribution location.

My curiosity compelled me to contact the Operations Director of Fairfax County Public Schools Food Distribution Services to learn more. She told me how the program worked and graciously agreed to provide me with meal distribution data when I asked if I could analyze it. Her data showed the number of meals provided in each distribution site over March - July 2020.

Along with that data, I was interested in analyzing how my county distributed meals to populations below the poverty line and impacted by COVID-19. In order to do this, I first sorted the data I was provided by zip code. I also accessed publicly available data on the number of COVID cases and populations below the poverty level.

I was initially unsure how to compare these numbers. Eventually, I played around with Tableau and decided that the figure below could allow me to compare regions across the county easily.

Additional information about the program: https://www.youtube.com/watch?v=uSikYd_6nlc

The figure I created was a map in which the darker a zip code is, the more meals distributed, the populations below the poverty level, and the COVID-19 cases there are. For example, the zip code 22306 is dark on all three maps, which means that because that area has a higher population below the poverty level and COVID-19 cases, more meals were distributed there. In contrast, an area like 22003 had high populations below the poverty level and COVID-19 cases, but fewer meals were distributed.

While this figure provided a lot of valuable information about the areas where meals were not adequately distributed, I realized that my new understanding of the materials taught in CS109 could allow me to conduct a more sophisticated analysis of the meal distribution data than my analysis from over four years ago.

I decided to revisit Tableau and make two graphs below, using the following assumptions:

My dependent variable (y-axis) would be the number of meals distributed in each zip code.
My independent variables (x-axis) would be the population below the poverty line and the number of COVID-19 cases in each zip code.

The graph I created above compares the population below the poverty level to meals distributed per zip code. In this graph, I added a linear regression line derived from the data—zip codes below the line are red because they are underserved compared to the other blue zip codes. 22306 has received over 200 thousand meals with around 4,000 people under the poverty line. 22033, which has nearly the same number of people under the poverty level as 22306, only received 75 thousand meals.

Suppose I wanted to know the number of meals needed to be distributed at a zip code to serve the community effectively. In that case, I can look at the prediction provided by the linear regression line. I think that there were dozens of ways to analyze this data, such as employing different machine-learning models. However, I chose linear regression for simplicity’s sake and because I hypothesized that the relationship between these variables would be linear.

Similar to the populations below the poverty line graph, the graph above compares the number of COVID-19 cases to the number of meals distributed per zip code. Zip codes like 22306 and 20170 have enough meals, while zip codes like 22003 need around 75,000 more meals to be served equitably.

In conclusion, the two new, more advanced graphs I made allowed me to understand that there were several zip codes like 22309, 22041, and 22003 that had lower meals distributed relative to their high poverty levels and COVID-19 cases, which is a strong indication that they are being underserved.

There may have been a lack of transportation to distribution sites in some zip codes. Or maybe the program's advertising method wasn’t as effective as planned. I think that this research is important because learning about why some zip codes were underserved could shed light on issues that could be fixed for future Fairfax County Public Schools programs.