Growing up in DC, the Smithsonian museums were a huge part of my childhood. They were always there, free, inspiring, and full of things that sparked my imagination. As I got older and more immersed in data work, I started thinking about how digital archives could be mapped back onto the physical world. With over 5 million open access pieces available online, where could people actually see these items in person?
In this project I aim to visualize the reach, content, and accessibility of the Smithsonian Institution’s Open Access collections through spatial mapping. Using Tableau Public, I created a series of maps that highlight patterns in object distribution, institutional focus, and thematic filters such as women-created works or historical time periods.
Data Collection Process
My curiosity led me to the Smithsonian’s Open Access API. I wanted to extract at least a million records to work with a strong sample, but I ran into limits. The API capped me at just over 600,000 observations. On top of that, the metadata was pretty complex—fields were deeply nested, and I had to write custom logic to pull information like creator names, object topics, and date ranges. For instance, I had to deduce race and gender directly from topic tags or creator descriptors. (Smithsonian API Pull Python Script)
The geographic side posed its own challenges. The open data didn’t come with coordinates or shapefiles, so I had to manually build one. That meant collecting addresses from Smithsonian websites and running the coordinates through R to create geographic points. (Clean Process Smithsonian Data R Script)
Visual Discovery & Design
My first map attempt was a bit of a struggle; I was aiming to build a proportional symbol map showing object counts per institution, but I couldn’t get the point sizing quite right. To make up for it, I leaned on color gradients and paired the map with a horizontal bar chart to make the count differences easier to see. It worked technically, but visually, it didn’t feel balanced or intuitive.
After resolving the sizing issue, I was able to construct a map with a more accurate proportional scale based on object count. But through user testing, I learned that the minimal base map made it difficult to recognize the location context, especially for people less familiar with DC. This feedback highlighted the importance of map readability, especially for public-facing tools. Other feedback I received, especially from curators, was to make it easier to filter by areas of interest like women’s art, Black art, or time periods.
So I created the above map, filtering for women-created works and updating the background to a more familiar street map. It helped with recognizability, but it made the map feel too busy. I realized I needed a compromise.
That led to the following maps: the proportional map and the pie chart map. These visuals keep the readable street base but fade it slightly so the data points pop. I also made sure the tooltips told a more complete story—each one includes the institution name, type, and a count of open access items. The proportional map also includes the most common object type, and the pie chart map shows how the pieces break down by time period, which I found particularly helpful for historical context.
Lastly, I made the reference map for users who might not know much about the Smithsonian network. It includes the full list of Smithsonian institutions by type, such as museums, libraries/archives, gardens, and more, with address info and website links right in the tooltip. Since the Smithsonian system spreads from New York to Virginia, but most locations cluster around the National Mall, I included a zoomed-in inset to make that area easier to explore.
My design approach was informed by both user feedback and my experience navigating museums. I wanted to create tools that were equally useful for curators, researchers, and everyday museum-goers.
Reflection
This is definitely a work in progress. There’s so much more I want to do. One big area I haven’t tackled yet is object type categorization. I pulled over 5,000 unique object types from the API. Grouping those into themes could really enrich the maps. For now, I have included the most common object type per institution in the tooltips, but this is just a start.
Another change I’d like to make is refining the symbol scaling. Some of the smaller institutions, like the Smithsonian Gardens or Environmental Research Center, disappear in comparison to the heavy-hitters like the Smithsonian Libraries. Adjusting the size scale or adding a minimum visual threshold would help with visibility.
I also really want to include images of the institutions in the reference map tooltips. A visual anchor would make it even easier for new users to connect with the spaces. And of course, I’d love to expand the filtered views, adding maps for Black art, Indigenous work, or specific types of media could support even richer storytelling.
What I’ve made so far is just a starting point. This project has been a way to bring my personal connection to the Smithsonian into conversation with my skills in data and design. And hopefully, it’s a tool that others, such as curators, tourists, or educators, can use to explore, discover, and appreciate all the pieces in place.
