I like my original mapping illustrations post, but it’s a little lightweight and I’ve belatedly noticed the requirement on the syllabus that we use a mapping platform. So here is take two.

I live in an area with a lot of film and television shooting activity. Our readings have me thinking about what it means for certain parts of the city to be shown often in mass media, while others may never appear at all.

I created a dashboard of maps in Tableau using the Film Permits dataset accessed through the NYC OpenData portal. The colored dots indicate the locations by zip code of shooting permits issued in the three-year period from 8/1/15 – 7/31/18 for commercials, documentaries, films, and television shows. Larger dots represent more activity.
(I can’t get the Tableau Public map to embed here, so please click through via the screenshot below to be able to access the informational rollovers.)

These maps show greater total number and diversity in location for film and television permits. I’m interested in how mapping can be used to show absence, so the map that is most interesting to me on this dashboard is the one showing shooting permits for documentaries. Over the past three years the majority of permitted documentary shooting activity has been in Manhattan and Brooklyn, with only a few projects in the Bronx and Queens, and only one in Staten Island. Less information and data of the documentary type are being created about the Bronx, Queens and Staten Island, which shapes how much presence they have both in current cultural awareness and how much will be available to people who wish to learn about these places in the future.

Some learnings:

• The raw data included multiple zipcodes for some permits in a single cell. I broke them apart into separate columns in Excel, then wrote a Python script that used Pandas and the melt function (method?) to reshape this into long, skinny data that Tableau could use to map each zip code separately. It would be better if I could have done the entire thing in Python, but I was under a time constraint and doing the first part in Excel was faster for me.
•  I’d destructively pared the dataset down to cover only three years by deleting the other rows in Excel. I should have left the data intact to leave myself the option of adjusting the timeframe using Tableau filters. Shooting permit data went back to 2012 in the original dataset. I’d like to map all available documentary permitting data to get an expanded view of which parts of the city are the topic of formal archival(?) content creation.

Things I need to learn how to do that would improve this dashboard: (1) include scale reference for each map, as the scale for the dots is not the same between them, and (2) synchronize the area shown in the maps to be identical.

A potential error with this project: I’m not sure whether shooting permits are also issued for shooting on permanent stage sets. I’ve inquired with someone who works in film and television, and I will update this post when I hear back.

Update: per my friend in the industry, these permits are not required for stage shooting unless there is substantial extra parking for trucks required, which happens often with television shoots. In those cases a permit is required, and so my mapped results may also reflect the locations of stage studios. To improve the focus of these maps, I could create a list of such places and add them as an information layer on the map.

This highlights the importance of collaboration. Any analysis one attempts to perform of an area with which they are not familiar will be inherently superficial. A map may tell a story, but it’s not necessarily a true story. It’s useful to solicit input from people who may have critical contextualizing knowledge, be able to identify missing or extraneous information, and can help provide an informed interpretation of results. My industry friend, for example, may draw entirely different conclusions from the same data and visualizations.

Summary takeaway from this exercise: identifying data is only one step. Visualizing a dataset does not automatically confer sufficient understanding of that data to construct a useful analysis.