readme.txt Food Access Vulnerability Dashboard
Streamlit Application | Master’s Project
This repository contains a Streamlit-based interactive dashboard developed to assess food access vulnerability using a multi-dimensional scoring approach. The project integrates socioeconomic, transportation, and geographic indicators to produce tract-level vulnerability insights.
The work reflects an applied, equity-centered analytic framework that moves beyond static “food desert” definitions by contextualizing access within local infrastructure, mobility, and structural conditions. All scoring matrices presented here are imperfect by design: they are starting points, not endpoints, and they are meant to be iterated, critiqued, and improved.
The USDA definition of a “food desert” was the catalyst for this project. At its core, that definition is binary, static, and driven primarily by income and proximity. While refinements have been made in academic contexts, these nuances have not meaningfully shaped real-world practice. The result is an insufficient and sometimes harmful operational definition.
I’m not a food access researcher by training. I’m a life scientist who focuses on health disparities. I did not set out to become “a data person”; I just wanted to understand and take responsibility for the statistical and methodological choices that underpin research. That commitment has led me here. My orientation now is simple: the life sciences caused harm by building and reifying biological essentialism into our tools. The only path toward repair is methodological rigor and explicit acknowledgment of past and present harms.
You’ll notice this project does not use demographic variables. That is intentional. I do not collect variables I am not studying, and I avoid demographic proxies in population-level modeling because they too often create false associations and obscure the mechanisms that actually drive inequity.
Developer’s Note: Iteration, Imperfection, and Critique Welcome
This project is intentionally submitted in its working state rather than as a polished final artifact. Analytic transparency, critique, and iterative refinement are core to its methodology. Food access modeling is imperfect, data are messy, and context-sensitive scoring systems evolve with community and expert feedback.
Even with excellent contributions from subject-matter experts, we run into the same conceptual wall: we measure “access” as proximity. Yet our lived experience of the social and environmental determinants of health does not comply with that binary quantitative frame. Upstream policy and downstream health outcomes bear the cost of this mismatch.
If you review or use this work, please critique it—methodological choices, weighting schemes, assumptions, and data pipelines are all open to examination. This represents an early iteration toward a more robust, community-grounded framework. Thoughtful critique is not only welcome; it is necessary.
I hope this serves as an Alpha, not an Omega. Consider this an invitation to build something better.
Limitations
Many: • My limited formal GIS and modeling training • Encountering Areal and Areal 2SFCA late in the process • Disruptions in public dataset availability • A climate where open methodological critique is increasingly fraught • The realities of working under academic timelines
Scope: This project focuses on a single county (n=8). This was intentional: better to fully analyze one county than expand prematurely with incomplete or low-quality data.
Contact
If you’d like to discuss the project, collaborate, or provide critique, you can reach me here: LinkedIn: http://www.linkedin.com/in/phinnmarkson-equity